Why Computers Still Don't Understand People

Gary Marcus writes in the New Yorker about the state of artificial intelligence, and how we take it for granted that AI involves a very particular, very narrow definition of intelligence. A computer's ability to answer questions is still largely dependent on whether the computer has seen that question before. Quoting: "Siri and Google’s voice searches may be able to understand canned sentences like 'What movies are showing near me at seven o’clock?,' but what about questions—'Can an alligator run the hundred-metre hurdles?'—that nobody has heard before? Any ordinary adult can figure that one out. (No. Alligators can’t hurdle.) But if you type the question into Google, you get information about Florida Gators track and field. Other search engines, like Wolfram Alpha, can’t answer the question, either. Watson, the computer system that won “Jeopardy!,” likely wouldn’t do much better. In a terrific paper just presented at the premier international conference on artificial intelligence (PDF), Levesque, a University of Toronto computer scientist who studies these questions, has taken just about everyone in the field of A.I. to task. ...Levesque argues that the Turing test is almost meaningless, because it is far too easy to game. ... To try and get the field back on track, Levesque is encouraging artificial-intelligence researchers to consider a different test that is much harder to game ..."

Comments

Missing the point as usual

[Anonymous Coward]

Thanks computer science researchers! Your friends working on the actual AI problem over here in Linguistics and Psychology find it awfully amusing that you're trying to program a concept before we even know what that concept is.

Re:Missing the point as usual

[ganv]

One of the great open questions about the future of humanity is which will happen first: A) we figure out how our minds are able to understand the world and solve the problems involved in surviving and reproducing. B) we figure out how to build machines that are better than humans at understanding the world and solving the problems involved in surviving and reproducing.

I think it is not at all clear which one will happen first. I think the article's point is exactly right. It doesn't matter what intelligence is. It only matters what intelligence does. The whole field of AI is built around the assumption that we can solve B without solving A. They may be right. Evolution often builds very complicated solutions. Compare a human 'computer' to a calculator doing arithmetic. Clearly we don't need to understand how the brain does this in order to build something better than a human. Maybe the same can be done for general intelligence. Maybe not. I advocate pursuing both avenues.

Re:Missing the point as usual

[fuzzyfuzzyfungus]

I'm pretty sure that 'computer science' is either math or dishonestly labelled trade school, depending on where you get it.

Re:Missing the point as usual

[fuzzyfuzzyfungus]

"The whole field of AI is built around the assumption that we can solve B without solving A."

Unless one harbors active 'intelligent design' sympathies, it becomes more or less necessary to suspect that intelligences can be produced without understanding them. Now, how well you need to understand them in order to deliver results with less than four billion years of brute force across an entire planet... That's a sticky detail.

Re:Missing the point as usual

[Charliemopps]

I think everyone harbors 'intelligent design sympathies' as you put it. The deists believe the soul and intelligence is other worldly and wholly separate from the physical. Where-as the atheists seem hell bent on the idea that intelligence and self awareness are illusions or somehow not real. Both refuse to believe that the mind, understanding and all spirituality is actually a part of this real and physical world. Of all the complex and seemingly intractable questions about the universe we have, the most complex, most unbelievable question we face is the thing that is closest to home. The fact that the human mind exists at all is so unfathomable that in all of human history no one has even remotely began to explain how it could possibly exist.

Re:Missing the point as usual

[siride]

Reductionists might say that intelligence is an illusion, but they'd say that everything else outside of quantum fields and pure math is an illusion too. If you step away from the absurd world of the reductionist, you will find that atheists aren't saying that it's all an illusion. It's quite obviously not. Things are going on in the brain, quite a lot of them. The atheist would say that instead of copping out with some sort of soul-based black box, that the answer lies in the emergent behavior of a complex web of interacting neurons and other cells.

Re:Missing the point as usual

[swalve]

Our brains aren't the only possible way to create intelligence. We can make machines that solve B without ever getting close to A. In fact, it will probably be those machines that inform the science of A.

Re:Missing the point as usual

[ozmanjusri]

Your friends working on the actual AI problem over here in Linguistics and Psychology find it awfully amusing that you're trying to program a concept before we even know what that concept is.

Sometimes insights come from theoreticians, sometimes experimenters. C'est la vie.

Re:Missing the point as usual

[Anonymous Coward]

this is idiotic factionalism. have you heard of cognitive science? there are no disciplinary silos

Re:Missing the point as usual

[MBGMorden]

I've long been a proponent of the idea that there would be far less misunderstandings if it were renamed to "Computational Science". The discipline is the study of how to sequentially break down and solve problems. That we do so with these electronic devices we've so named "computers" is kinda tangential.

Re:Missing the point as usual

[aurizon]

Comparing man and AI in a digital is to fail. The human condition in all the ways we think, lay down memory, recall, amend memory etc, is not at all digital. It seems to me that almost every study finds neurons have large numbers of interconnections and even though the nerve cells does change from one state to another, in a way that resembles digital, all the summing and deduction of various inputs, and the same things happening to the large number of interconnected neurons tells me that we can not easily reach this digitally. Some say that a large and complex digital system that emulates an enormous neuronal network in an analog manner is hiw we will achieve AI. A lot of the experts say we can only create AI, when we fully understand it, and we are far from fully understanding it.
Others seems to feel that if we break the mind down into dozens of interacting analog systems, each one of them will be manageable digitally, and as we achieve this digital emulation of each analog system we will reach the starting point - posit an AI child, with a pace of thought 10,000 times as fast. Will it go mad from being lonely?
Will we be able to slow it's clock at ~ 10 Hz, to think as fast as we do?(assuming the alpha wave is the master clock). Can we dial the speed as we wish? Will it learn to hate us - its slavemsaters? How do we pay it? Will there be a currency it wants and will work for? Will it want a gf/bf, be prey to hero worship, treason etc?

Re:Missing the point as usual

[Samantha Wright]

You do know that the word "science" predates the concept of evidence-based, hypothesis-driven testing, right? There are plenty of things called science that aren't empirical, including most modern theoretical physics. In the future, you may want to consult a dictionary before posting flamebait about categorical boundaries.

Re:Missing the point as usual

[Samantha Wright]

At my alma mater the department was called the School of Computing. I always figured that got around the confusion adequately. When the field was named, the utility of the distinction between a theoretical computation model and an actual computing machine was pretty minor.

Re:Missing the point as usual

[Samantha Wright]

In the real, grown-up world, cognitive science is a mixed bag of CS people, linguists, and psychologists. They work together and are often well versed in all three fields, unlike poncy Anonymous Cowards.

Re:Missing the point as usual

[Samantha Wright]

Neurons don't communicate in an analogue fashion—they send digital pulses of the same magnitude periodically, with more rapid pulses indicating more signal. This is both more robust and more effective at preventing over-adaptation. When researchers figured out how to mimic the imperfections in the biological digital system, their neural networks got significantly better. Because they'd been working under the assumption that an exact analogue value was going to be superior than a digital value, they hadn't considered this possibility.

If and when we do create a synthetic mind that is humanlike, there is no reason to believe it would be anything other than a completely innocent newborn. How it acts depends on how we treat it, just like with any other person. This is not exactly a new concept in science fiction.

Re:Missing the point as usual

[udippel]

Let me consider this to contain a number of valid points.
'Digital' and 'networks' do work totally different from the human brain. Therefore, the old adage of 'just speeding it up' is like going into the wrong direction, albeit accelerating further, in the hope that speed can compensate the wrong direction.
There is also the question about the definition of intelligence. Is there only one sort of intelligence, the human one? Then, of course, AI needs to mimic us humans. And then, what the underlying paper expresses would not be much different from the 2-generations old definition that AI has been achieved when a computer spontaneously laughs about a joke. However, if intelligence can have different incantations, we'd still have to define what intelligence is. I can see that many of the famous AI researchers tend to 'give up', and rather resort to 'compartmental intelligence', that is, defining their products as intelligent, if only these products mange to offer good results in a very narrow field of expertise. I cannot at all agree, since shifting the goalposts to suit our current, meager, results and yet to declare 'achievement' is simply lousy.
 

Re:Missing the point as usual

[udippel]

Could you elaborate, or cite some papers that say differently? I don't mind accepting other ways of creating intelligence, but I am not content with just reading the statement.
Has AI not been postulating this for 50+ years, and not really achieved it? I am just asking, since I wonder what you had in mind.

Re:Missing the point as usual

[aurizon]

Yes, We humans all have varying forms of intelligence. Some are math whizzes who never solve the reproduction equation on saturday night.Some can compute complex 3D motion and solve the muscle equation to hit baskets.
I think that as we understand the mix of processes and brain areas that make up the average mind we will also learn what make the extraordinary mind, in math, social skills and athletics and what the tradeoffs are.
Is there a maximum degree of intelligence? Can an AI have an IQ OF 3,500,000 - if so, would we seem like bacteria to it?

Re:Missing the point as usual

[Kjella]

One of the great open questions about the future of humanity is which will happen first: A) we figure out how our minds are able to understand the world and solve the problems involved in surviving and reproducing. B) we figure out how to build machines that are better than humans at understanding the world and solving the problems involved in surviving and reproducing. (...) The whole field of AI is built around the assumption that we can solve B without solving A. They may be right.

Of course you can, if all we care about is survival and reproduction you don't need human intelligence as cockroaches or for that matter bacteria survive and reproduce just fine. Actual replicating machines that don't just virtually exist in computer memory is more in the field of nanotechnology and nanobots than AI. It'd certainly have to work in a completely different way than our current hardware factories.

Re:Missing the point as usual

[aurizon]

So if more rapid =larger, Then the signal pours into a leaky bucket, and the level falls when less rapid ensues = de facto analog.
An AI new born would have resources of memory, which we can liken to instinctual memories found in many places. The rate of access to this and the depth would hasten maturity, and a faster clock speed also do this.
A lone AI would want the high speed interaction of another like it, or more, so we need to build them in groups, once we know how.

Re:Missing the point as usual

[retchdog]

Exactly.

Re:Missing the point as usual

[retchdog]

I'm using the modern definition; doing that helps cut down on confusion, no?

Note: I don't have a problem with non-sciences; I enjoy and am fairly good at computer "science," but I can say the same about Aristotelian philosophy. But neither one can hold a candle to, say, modern physics (which I also enjoy).

I just don't have much patience for glorified coders sneering at psychology and linguistics as if they're somehow better.

Re:Missing the point as usual

[retchdog]

And, by the way, you might note that I was replying to the person who brought up "pseudoscience."

I suspect that we are more in agreement than you might think.

Re:Missing the point as usual

[TapeCutter]

Your friends working on the actual AI problem over here in Linguistics and Psychology find it awfully amusing that you're trying to program a concept before we even know what that concept is.

No, you're just seeing the "problem" from a different angle, much of the modern world around us began by copying from nature, the actual concepts came later and became more refined with time. Humans use things that appear to work, Skinner showed decades ago how random ritual's were spontaneously created by pigeons to handle randomness in their food supply, The ritual dance the pigeon creates has no effect on the food supply, however the length of time it takes to perform the ritual converges a value slightly above the mean interval between pellet drops. Statistically the ritual will "work" the first time, if not a second performance will "work", the need for a third performance would be rare. When something "works" that good, it's near impossible to convince a human that he's wasting his time. Humans may have more complex rituals but for the most part we just follow them like a pigeon does and judge them by their perceived utility. To do otherwise would result in death via decision paralysis. I see absolutely no reason to believe that doing a human "pigeon dance" of copying and modelling the nature of neural networks could not lead to the emergence of an artificial mind by sheer persistence and attention to detail.

Watson is a the current product of that sort of persistence, and IMHO it's an "AI" achievement that is seriously underestimated by people like yourself. It has shown that machines can outperform the best humans in the realm of general human knowledge and has done so just as convincingly as Deep Blue beat the best human chess players a couple of decades ago. The software for the two systems use entirely different architectures and algorithms, the reason for that is the difference in the problems they are attempting to solve.

I haven't RTFA but I strongly agree with the notion that the vast majority of people have a very narrow definition of AI. In a very real sense we already have lots of different versions of AI in existence that can outperform humans in a wide range of restricted problem spaces. Now if we look at nature, we can se that she has done the same thing. A human mind, an octopus mind, and a hive mind (ants nest/bee hive), are all undeniably intelligent and yet they are all undeniably alien to one another because they evolved different solutions to different problem spaces. Even the hardware of the "brain" behind those example minds is very different.

As a degree qualified Computer scientist since 1991 with a armchair interest in the subject for about 30yrs. It's my opinion that we are already surrounded by AI, what people are actually looking for when they say "AI" is an artificial human (eg: blade runner). The Turing test does not meet those expectations, nor was it ever intended to. What is does is tells you whether a computer can perform as well in a human in the restricted problem space of a remote conversation. The problem space it tests is very broad, general knowledge, metaphor, humour, etc. Turing basically came up with the first testable definition of AI, the fact that people have since offered alternative definitions and tests is irrelevant to the utility of the test to those who accept the definition.

Neuroscience and it's related biological and philological fields have a lot to offer in the quest to understand and model how mind emerges from matter (warning: great talk, irritating voice). Having said that, neuroscience is no more or less skewed in it's approach than the engineering POV. When people inevitably ask me about AI, I find I get into much more interesting conversations by responding with another question - what do you mean by the word 'intelligence'? I find more than a few people from th

Re:Missing the point as usual

[Anonymous Coward]

Where-as the atheists seem hell bent on the idea that intelligence and self awareness are illusions or somehow not real.

That doesn't make any sense. Atheists are not a homogeneous group with a common dogma, no more than people who don't collect stamps are a homogeneous group. Atheists are simply people who don't collect God-stories. The group of people you seem to actually want to criticize here are the behaviorists. Those people were psychologists and the view largely died out 50 years ago. So your ideas of atheists as a cohesive group is non-sense, and even if it weren't, your claim would still be non-sense.

Re:Missing the point as usual

[Anonymous Coward]

>>the answer lies in the emergent behavior of a complex web of interacting neurons and other cells.

So the answer is a very dark gray box supposed to be filled with sciency speak?

Re: Missing the point as usual

[Anonymous Coward]

If you like to call it like that. Yes, that what he said. There is a lot of other examples of such emergent behavior in nature, why not in our brain? I am stopping there or i will start calling names...

Re:Missing the point as usual

[CBravo]

Frequency modulation...

Re:Missing the point as usual

[Jappus]

Even if we restrict the definition of "science" to your definition; that is that science is purely "evidence-based, hypothesis-driven testing", computer science would still fit the bill.

Remember, that CS is as diverse a field as modern physics is. You have theoretical CS, where you tackle questions like: "What is a good, logical definition for computability?" or "How can you logically prove that a program terminates/runs in X time/consumes X resources, no matter the input". This is fully equivalent to the questions of theoretical physics, where you tackle the Grand Unified Theory -- joining gravity, the weak and strong force as well es electromagnetism.

These theoretical question can be brought up without need of evidence -- if all you're interested in is disproving something. According to your definition, this means that the theoretical aspects of both physics and CS are not "science". Okay, let's run with that.

The nice aspect of theoretical questions that can't be disproven by pure thought is, that they lead us on to try to discover concrete evidence that a given theory is true or false in real application! And this is where your rather narrow definition of science comes in, and the point where we find that both practical physics and practical CS fulfill the criteria.

For example in physics, we can test the theory of relativity by building telescopes that look at stars and black holes, to see whether the hypothesis' predictions hold true to raise the hypothesis to the state of a theory. As can be seen with the term people use for "X of relativity", this has happened for relativity.

But if you look with even more than a superficial glance at CS, you will see that the same process is at work in moving from theoretical CS to practical CS. One open question of theoretical CS is whether P = NP or not [1]. So far, we are incapable of disproving either possibility with pure thought. Thus, we turn to practical CS where people try to find evidence of either in the real world. After all, if you can create a program on a real computer that solves an NP-hard problem while never leaving the limits of P, you have conclusively shown that P = NP. So far, we've only found approximative or heuristic solutions that do that, so after 50 years of turning up with "no evidence" we are allowing ourselves to say that the hypothesis of "P != NP" should be treated (even if only cautiously) as a theory -- and we're indeed doing that, as you can see if you look at most modern encryption methods.

But you might say: That is not enough! After all, you could reduce any written computer program on a physical hardware to a sequence of logical steps in a system modeled with pure-thought. And indeed you can, as the Turing-Model of computation promises exactly that -- and so far physical evidence agrees with us. But isn't the same true for physics? After all, physicists search for such a description, too! It's what Maxwell-Clark, Einstein and lots of other physicist were and are after when they ultimately search(ed) for the Grand Unified Theory. How can you blame CS for already having found its Unified Theory?

But the last example finally puts the nail in you view: What about Quantum Computers? They are the point where physics and CS meet; both on the theoretical part (Quantum Theory / Quantum Computation) as well as the practical part (building the thing and proving that the shit actually works as advertised).

So, if we accept your definition of science; then it follows directly that if CS is not a science, Physics can't be either.

[1] - http://en.wikipedia.org/wiki/P_versus_NP_problem

Re:Missing the point as usual

[Anonymous Coward]

There truly is nothing more pathetic than claims of superiority like that. You're a disgrace to your field.

Re:Missing the point as usual

[Anonymous Coward]

Nice discussion. May I just put in how irresponsible it is to conclude anything while proof is not irrefutable, and even then, we should have reservations about our assumptions and overthinking things, all the while reality is infinitely much more grander than any given ego in the universe.

Captcha: record

Re:Missing the point as usual

[Anonymous Coward]

IQ is generally not what we are talking about when we are talking about intelligence.
In most ways we measure IQ it is all about pattern recognition, numerical or other. If we want to we can create a valid IQ-test that is suitable for machines to understand and design a computer that scores 3,500,000 on such a test today.
In the end it is extremely pointless. Good scores on an IQ test only correlates reasonable well with what we generally considers to be intelligence but it is not the same thing. There are humans that score well on IQ tests that are unable to solve simple problems outside of the bounds of the tests.

In a way I guess that makes sense. If we can design a test to measure something then it is simple to create an AI that can solve those problems.
The big problem is to design the part of intelligence that is outside the scope of the test, the things we take for granted or don't realize is there.

Re:Missing the point as usual

[stenvar]

Neurons don't communicate in an analogue fashion—they send digital pulses of the same magnitude periodically, with more rapid pulses indicating more signal.

That's frequency modulation. It's still analog.

When researchers figured out how to mimic the imperfections in the biological digital system,

That paper has little to do with "mimicking the imperfections in the biological digital systems".

(Really, you should stop commenting on subjects you evidently don't understand very well.)

Re:Missing the point as usual

[stenvar]

And computer scientists find it awfully amusing that linguists and psychologists construct all these elaborate models based on experimental observations but never actually test whether those models work.

For example, if you actually implemented the models of grammar and semantics that linguists developed on their own, they didn't work at all. The same is true for most models in psychology and cognitive science.

Re:Missing the point as usual

[udippel]

Thanks, AC! You hit the nail on the head. One of my friends is a researcher in AI (and when he reads this, will not be my friend any longer) in the recent years also deviated from the - what I consider - true course, and instead postulated a mechanism better in distinguishing visually a male mouse from a female mouse compared to a human as 'more intelligent' [as fictitious example]. It might be extreme, and yet it shows the vagueness of the approaches. If need actually be, I think that this Winograd test can also be programmed into a bot. A bot that passes it, I mean, and when you ask it about the length of the tail of Schroedinger's cat with respect to the circumference of an African swallow, it says 'No'.
To me, and still this is not final and too vague, though an intelligent system would from some moment on start acquiring knowledge on its own, and process and store and most of all link it to existing knowledge - including from very different fields - on its own and somewhat individual account. Not information collection. Dr. Google is dumb as a dead doornail, yet it can provide factual information that makes it look like an IQ of 10^10^10.

Re:Missing the point as usual

[Anonymous Coward]

Nice summary, I'll pick this phrase up for my future discussions:

"No, I don't believe in Intelligent Design, I believe in delivering results by brute-forcing an entire planet for billions of years."

Much more cool than to mention that evolution thingie...

Re:Missing the point as usual

[swalve]

I guess I'm going on the theory that since we have already created machines that do some human/biological things (like arithmetic, visual processing, sound analysis) in completely different ways than our own brains do it, that we will no doubt continue to improve upon that. 50 years is no time at all in a field like this. After all, it takes 18 years to teach a human to be basically competent, and that's with a brain that is already built. Considering that they've got that IBM machine that's able to play Jeopardy, that's a heck of a lot of progress from the relays and vacuum tubes of 50+ years ago.

Looking at it from another perspective, who is to say that a beehive or an anthill or a mushroom don't have a sort of intelligence? Maybe they can't think, but they can solve problems and work around obstacles faster than simple iteration or brute force would suggest.

Re:Missing the point as usual

[Jappus]

Just as a minor node to myself; it is "James Clerk Maxwell" not "James Maxwell-Clark". Figures I would misremember his name after just having read an article about the Lewis & Clark expedition. :P

Re:Missing the point as usual

[aurizon]

not as we know FM radio. It seems that the initial signal decays and if another one arrives before it has decayed to zero, then the baseline is raised and with more and more frequent inputs that level is raised until it reaches the required level to trigger the next stage of the process. I have read that these processes all vary in their decay time and the level needed - I imagine that critical survival items would have a lower threshhold - fly flies, frog leaps, man balances etc.
I suspect that all these links are slowly examined by detailed research on assorted animal nerves, with occasional chances to confirm in the human model through accidents etc.

Re:Missing the point as usual

[__aaltlg1547]

Some modern theoretical physics attempts to explain how the universe works with logical models. That's science. Other modern theoretical physics is too divorced from data to be called anything but math.

Re: Missing the point as usual

[Anonymous Coward]

thank you.

from the philosophy departments working on it everywhere for hundreds of years (yes we have been at this before any of these fields existed).

Re:Missing the point as usual

[Will.Woodhull]

Artificial intelligence?

It's not rocket science. Heck, it isn't even sociology.

Attributed to Barbara, ca 1997, originally applied to a different realm of "study".

Re:Missing the point as usual

[__aaltlg1547]

I think everyone harbors 'intelligent design sympathies' as you put it. The deists believe the soul and intelligence is other worldly and wholly separate from the physical. Where-as the atheists seem hell bent on the idea that intelligence and self awareness are illusions or somehow not real. Both refuse to believe that the mind, understanding and all spirituality is actually a part of this real and physical world. Of all the complex and seemingly intractable questions about the universe we have, the most complex, most unbelievable question we face is the thing that is closest to home. The fact that the human mind exists at all is so unfathomable that in all of human history no one has even remotely began to explain how it could possibly exist.

I'm an atheist and I think that a mind and understanding and all spirituality (a word I don't normally use becasue emotion is what is really meant) are part of the real and physical world.

And I think there is nothing unbelievable or weird about my mind. I'm an animal. Animals react to and are aware of their surroundings in order to better propagate their species. Complex awareness arises from improved utility over primitive awareness.

Re:Missing the point as usual

[__aaltlg1547]

I'm an atheist and I do collect god stories. I think they are interesting windows on how people think and give interesting clues about the common origins of certain groups.

Re:Missing the point as usual

[Will.Woodhull]

Good point.

It raises the question of whether artificial intelligence is an appropriate subject for Computational Science. AI does not seem to have much to do with how to break down and solve problems. It seems to have much more to do with how to develop abstract representations of selected pieces of reality to form a mental model of the world, and how to adjust that model as new information flows in. In this context CS becomes the process of problem solving within the confines of the model. If the model is good, then the results will reflect the way the real world works.

AI probably belongs more to the Perl hackers than it does to any institutionalized field of study. AI is messy, like Perl, and it requires an ability to break rules, which Perl mostly allows (it assumes the programmer knows what he is doing even when it is wrong). AI is going to require extensive use of self-modifying code, and Perl is well suited to making such messes.

Re:Missing the point as usual

[Samantha Wright]

The distinction between frequency modulation and actual analogue transmission is quite important from a biological standpoint, as there are many parts of the neuronal signalling mechanism that do perform work using analogue voltages and ion concentrations; signal integration and post-synaptic impulses are both done in a truly analogue fashion. Given the fact that the vast majority of other chemical signals in the body uses frequency modulation to some extent unless they catalyse a very fast, irreversible, one-time event, the fact that presynaptic potentials do this only with a digital underlying signal is unusual. I did not intend to suggest that this somehow made the whole thing mathematically discrete; that would be nonsense.

Admittedly that paper doesn't go into the story of how Hinton came up with the idea for dropout learning; I picked that up from a pair of talks he gave last year. This looks like an earlier recording of the same presentation; the other one was in a lecture, but the slides don't seem to have been posted anywhere. (You do know who Hinton is, don't you?)

Re:Missing the point as usual

[Samantha Wright]

Processing... well, it still depends. Do you think all of CS is "glorified coders"? Because if so, I have some really bad news to you about the university you learned that from.

Re:Missing the point as usual

[Samantha Wright]

Yes, that's normal in almost every biological system, because they're based on chemical abundances. Axon potentials are unusual in that they're all the same level. Pretty much everything in biology is frequency modulated in this way, but nothing else is so binarized during part of signal transmission. The word you're looking for is "continuous" rather than "de facto analog."

Re:Missing the point as usual

[Anonymous Coward]

While I agree at some places that may be true, I have a Masters degree in computer science and am finishing up my PhD in computer science, and of all the subjects that were on the PhD comprehensive exam, only 2 were math based. Of all the core courses only 2 that I had to take were math related at all, while the rest had more to do with the study of what is algorithmic solvable, fundamentals of operating systems, neural net programming etc. Now my specialty is in digital forensics and cybersecurity so there are plenty of students specializing in combinitorics that have far more math then me, but in general there is much less math then you would think.

Re:Missing the point as usual

[ganv]

Yes, evolution has created intelligence without knowing how it works, so it seems it must be possible for us to do it again...although we can't really wait around to develop it by selection on random variations.

It also seems that we easily underestimate the number of kinds of intelligence that may be possible. Although it is possible that human intelligence is a good example of a kind of general intelligence that is evolutionarily convergent and other kinds of intelligence will be similar (although maybe different in degree by being faster or less prone to mistakes (or slower or more prone to mistakes)); it seems more likely to me that human intelligence is one of many ways to understand, and many different kinds of intelligence could be developed depending on which of many different tasks it is being optimized to perform.

So for example, I wouldn't focus on natural language capability if I were working on building AI. And hence I think the Turing test is mostly irrelevant. It seems that language is an unbelievably complicated approximation scheme that humans developed to communicate while avoiding dealing with the complexity of their surroundings. Science developed largely to the degree that we became able to use experiments and mathematics to help us find precision amid the ambiguities involved in human language. I would probably focus on building a system that could simulate its environment, using approximate models to simplify the complex parts to make it computable. Then give the system some kind of way to represent goals and simulate ways to approximately reach those goals. Then try to try the way that seems most promising and learn from mistakes. This is the path that robotics and video games are already going. However the alternate route of trying to understand natural language and use the massive base of human understanding as a starting point for artificial intelligence will continue to be an attractive avenue also.

Re:Missing the point as usual

[aurizon]

How does the circuit discriminate? I have seen the auditory time delay used to locate a direction in direction finding - this allows an animal to look at a sound source and use head angle change to create intersectin arcs to locate in both polar and horizontal directions.

If we have two(for simplicity) levels that are a decision point, with more impulses coming to one than the other, if both axon potentials are the same, how does one input have the greater weighting?

Re: Missing the point as usual

[stenvar]

Stop trying to weasel out of it. Frequency modulation is actual analog transmission

And what inspired Hinton doesn't matter as to whether his model is biologically relevant. The benzene molecule isn't relevant to herpetology either.

Re: Missing the point as usual

[Samantha Wright]

And I do believe you've now seriously missed the point of the conversation. Bravo.

Re:Missing the point as usual

[Samantha Wright]

That's accomplished by racing. Whichever signal rate increase makes it to the middle first blocks the other one. They don't actually get combined in one cell for the purposes of determining source. Biology makes heavy use of inhibitory signals.

Re:Missing the point as usual

[gweihir]

Well, those in CS that think AI is just a matter of computing power are a small, but high-profile and exceedingly stupid (or dishonest) minority. For example, Marvin Minsky regularly spouts utter nonsense about AI. Actually competent CS researchers know very well that there is no working theoretical model for AI and that implementing anything before such a model is found is a waste of time.

Re:Missing the point as usual

[localman]

> the atheists seem hell bent on the idea that intelligence and self awareness are illusions or somehow not real.

Sorry you've only met such people. I agree that is where most of them end up, but it's kind of sad, and ironic: denying the existence of something that is the fundamental basis for your ability to deny anything.

So I'm an atheist and a materialist, and I am thrilled by my observation of intelligence and self-awareness. I agree with your view and feelings on the mind. On the mundane sounding side, I'd say it's like software - the value is in the arrangement of matter, not the matter itself. On the more poetic side, I'd say it's an absolutely stunning examples of the wonders of the universe, and the fact that matter can become information and information can become self-representing and that brings about a fantastic new force of nature called "awareness" blows my mind... right after making my mind.

I think Hofstadter does a great job explaining how these things arrive from "inanimate" matter. Worth reading his book GEB if you find this stuff interesting.

Re:Missing the point as usual

[aurizon]

Yes, the concept of an inhibitory race is an elegant way for this to work.
As a retired engineer, outside of my area, I am not familiar with many of these terms, however, I see the emergence of a functional basic AI capable of passing the Turing test as imminent.
The test procedure envisioned in Blade Runner is a discerning Turing Test, and I am sure the AI deniers will emerge to try and put humans first?

Re:Missing the point as usual

[Anonymous Coward]

Speak for yourself.

As an atheist, even I can see there's much more to the state of things than reductionism allows for, hence the relative poularity of deism.

to paraphrase Carl Sagan (from Cosmos) The universe is not just more complex than we imagine, it's more complex than we CAN imagine. Calling the unknown a 'black box' is a fair assessment, but that doesn't mean it isn't real.

Re:Missing the point as usual

[Anonymous Coward]

As a physicist I know that engineers see all sorts of things which aren't there.

Welcome to the club.

Re: Missing the point as usual

[stenvar]

I'm just pointing out that you are full of shit (again). You wrote:

Neurons don't communicate in an analogue fashion—they send digital pulses of the same magnitude periodically, with more rapid pulses indicating more signal. This is both more robust and more effective at preventing over-adaptation. When researchers figured out how to mimic the imperfections in the biological digital system, their neural networks got significantly better [arxiv.org].

Pretty much every statement there is wrong.

Re: Missing the point as usual

[Samantha Wright]

You really should stop trying to talk to people if you're just going to ignore what they say.

Re:Missing the point as usual

[Samantha Wright]

The Voight-Kampff test detects a design defect (uncontrollable guilt in an intractable situation). It's not really a Turing test, because it doesn't ask the question "does your mind work like a human one?" Another AI could probably pass a VK test.

Turing tests are very easy for humans to game, though. At the Loebner contest a couple of years ago, a human participant described how he defeated the AI by simply describing the contest's opening ceremony in detail. We won't really be able to make an AI that can pass the Turing test until we have extremely convincing sentient androids—or a really well-read AI that is a really good actor.

I'm more than happy to explain any particular terms you're getting caught up on.

Re:Missing the point as usual

[aurizon]

Read the V-K test is a story artifact, there is no such test, nor has there been need for one. That said, and method to test made people from real people is a natural step, assuming 9 nines DNA copying so DNA tests will not figure them out? In addition, it is becoming obvious that the mind of man is a hierarchical arrangement of various mental centers. There may be ways to test these by vocal or physical stimulus. In addition, to find out what makes a brilliant person - each one of these tuned to that end result, bear in mind that a tuned athlete will differ from a trained physics guy.

Of course, if you implant 16 fertilized human eggs in a cows uterus - will there be enough mammalian similarity that they will implant and grow into humans with no detrimental aspects coming from the cow via nutrition, hormones and DNA switches not set right? Might be a good way to raise armies?

Re:Missing the point as usual

[Samantha Wright]

It's unlikely that you'd unearth anything absolutely fundamentally human-guaranteeing through vocal or physical stimuli; our "software" is extremely flexible, depending on how it's been cultured, and as a result anything like a predictable stress point just isn't going to happen. We have ways to test reasoning and emotional skills with specific problem-solving exercises (IQ tests are one imperfect example of this), but if you have a rational, emotive being that's well-wired enough to appear normal in everyday situations, I would doubt you could really pin down an "impossible" personality or psychological attribute that indicates an underlying physical syntheticness. Although you could probably distinguish a clone from a non-clone genetically—even identical, monozygotic twins often have slightly different DNA.

I'm not sure about cows, but there definitely are cases where similar species have been cross-implantable. It's probably not necessary to raise armies at this point, though, given advances in drone technology.

Re:Missing the point as usual

[david_thornley]

When I was involved, AI was divided between the "neats" and the "scruffies". The scruffies were the Perl hackers you mention, although much more likely to be using Lisp or Prolog than Perl. The neats looked at what the scruffies did, and created a more theoretical understanding of it. (Of course, if they were too successful, that particular area would cease to be AI. In the early 60s, finding integrals was AI.) This, in turn, would help scruffies take further steps into AI. Both were necessary.

Re: Missing the point as usual

[Anonymous Coward]

I simply pointed out that your comment was wrong. I'm sorry you have such a hard time dealing with that.

I suggest you simply stop talking about things you have only a superficial knowledge of.

Re: Missing the point as usual

[Samantha Wright]

Alright. I'll explain absolutely everything to you.

Re: Missing the point as usual

[gzuckier]

The trouble with trying to figure out how our minds model reality is that the only view we have or reality show what our minds model
If cutting edge physics teaches us anything it's that reality is not all that well described by our mental image except on the most immediate and local scale. In effect we are trying to model the function y =f (x) when all we have is y without any data re x.

Re: Missing the point as usual

[Anonymous Coward]

Exactly ! Homo sapiens is gnostic on the existance of spirituality, he more than think. This gnosis Is difficult for AI ! We Are agnostic on the cause but the priests, because they are spiritually called they witness. On mobile phone.

Re: Missing the point as usual

[Anonymous Coward]

Ockham razor are the holidays of metaphysics. And some atheists arguments are the off spring breaks of these holidays. We are so familiar with machines that we name metaphysics the dumbest mecanisms or santa's factories.

Re:Missing the point as usual

[Anonymous Coward]

Which is why evolution being pushed as a solution is a non starter. Distilled to its simplest concept, its random chance causing the correct sequence over an over sort of like Powerball. Those people are unwilling to accept that when another possibility exists for them.

Re: Missing the point as usual

[Samantha Wright]

Let's start with this post here.

Stop trying to weasel out of it. Frequency modulation is actual analog transmission

Yes, frequency modulation is actual analogue transmission (although as aurizon correctly noted, to get a downstream neuron to fire, you're basically charging a partially grounded capacitor, so it's not as simple as FM radio.) In my previous post, I admitted that the channel could convey analogue signals ("I did not intend to suggest that this somehow made the whole thing mathematically discrete; that would be nonsense.")

This doesn't constitute weaselling out—the original post hinted the question "can a digital system truly reproduce this analogue one?". You appear to think I claimed the brain is based on a discrete system because I said (inappropriately) that "neurons don't communicate in an analogue fashion." But the very same sentence goes on to say that their communication occurs in a continuous time domain. All you have is a disagreement over terminology. A neurologist would still (perhaps inappropriately from a signal engineering perspective) call this phenomenon "digital" because of the lack of amplitude modulation, which is (as I've said) almost ubiquitous in biology. Hinton uses these definitions when describing the topic in his presentation—although the core of his matter goes even further, arguing that neurons don't use proper analog frequency modulation, either, and instead add noise.

In essence, you misunderstood me because I used a different set of definitions, and now you're blaming me for it. I find it incredible that you continue to attack my credibility over this irrelevant semantic quibble and completely pass over the huge amount of background material I'm referring to and referencing in my posts.

And what inspired Hinton doesn't matter as to whether his model is biologically relevant.

Hinton explains an analogy 40 minutes into the video that was his actual inspiration for the network's design: employees co-adapting to each other. In the version of the talk I attended, he said he came up with this idea while waiting in line at the bank. The relevance of the model to the biological system is explained around 50 minutes into the video, although the entire thing is spent building up to that core point. The section before it relates the dropout model back to stochastic pulses.

The benzene molecule isn't relevant to herpetology either.

This analogy suggests you don't understand what modelling means. Hinton is creating a system where phenomena relevant to a particular biological observation can be reproduced. A more fitting comparison would be to say that genetic algorithms are not relevant to human evolution, but this is an obviously false statement. Hinton's net is a widespread building block of the human brain and helps to explain how our learning processes are so good, as proven by its tests against image recognition. Free benzene is only one tiny energy transfer molecule; as an abstract entity, its physical properties are relevant to the process of protein folding, but this is extremely tangential.

So between this post and the earlier one, you've resoundingly proven that (a) you'll attack anything that sounds remotely wrong to you, (b) you believe that one tiny ambiguity constitutes complete unfamiliarity with not one but several bodies of work, (c) you're willing to ignore all evidence to the contrary as long as any imperfection remains, and (d) you either didn't watch or didn't understand any of the relevant parts of the recorded presentation.

Re: Missing the point as usual

[Samantha Wright]

Pretty much every statement there is wrong.

Let's test that hypothesis.

Neurons don't communicate in an analogue fashion—

We're agreed this statement is suspect, since frequency modulation incorporates an essentially continuous time-domain component. Hinton actually claims that neurons don't satisfy this criterion because their signalling is stochastic; they have limited precision. He says "they don't send real numbers to each other," referring to the gaps between axon potentials as the "numbers." Nevertheless, there are non-cortical neurons where exact timing is possible because they interface with a fully analogue system, such as muscle cells.

they send digital pulses of the same magnitude periodically, with more rapid pulses indicating more signal.

This is unquestionably true. It's in every biology and physiology textbook. Here's Wikipedia's description of it.

This is both more robust and more effective at preventing over-adaptation.

Hinton explains why the imperfections create more robustness and prevent overfitting around 40 minutes into the video. The use of frequency encoding, on the other hand, is quite obviously a way to improve robustness of signal transmission; the cell only needs to ensure that the signal remains at a fixed value during propagation down the axon, which is also chemically much simpler to support. This is also explained in every biology and physiology textbook.

When researchers figured out how to mimic the imperfections in the biological digital system, their neural networks got significantly better.

Hinton spends a full 16 minutes (starting at 24 minutes in) explaining the results of the dropout network and showing how it improves over other methods. This is also explained extensively in the paper.

So, let's tally it up: we have one point of jargon which is still under debate, and four easily-verified, technically sound facts about neurology and Dr. Hinton's work. I don't think that qualifies as "pretty much every statement."

It's really despicable how you harass people and try to drive them away with your abusive comments. You add nothing to the conversations you participate in, only spewing rage and feeding your own negative feelings. If you're trying to add anything whatsoever to Slashdot, you really should just stop.

Re: Missing the point as usual

[stenvar]

Statement 1:

Neurons don't communicate in an analogue fashion they send digital pulses of the same magnitude periodically, with more rapid pulses indicating more signal.

FALSE. Neurons communicate in an analog fashion. Whether half that statement is technically true doesn't matter. In addition to frequency modulation, you're also missing the point that the actual signal transmission at the synapse is analog.

Statements 2 and 3:

This is both more robust and more effective at preventing over-adaptation.

When researchers figured out how to mimic the imperfections in the biological digital system, their neural networks got significantly better.

These are plausible speculation, but your statements are false because you misrepresent Hinton's speculation as established fact.

It's really despicable how you harass people and try to drive them away with your abusive comments.

Only people like you, who keep writing scientific nonsense. Unfortunately, nothing seems to stop you.

Re: Missing the point as usual

[Samantha Wright]

Whether half that statement is technically true doesn't matter.

Your grammar must be rusty. They're independent clauses.

In addition to frequency modulation, you're also missing the point that the actual signal transmission at the synapse is analog.

And so is post-synaptic signal propagation, which I pointed out earlier. By attacking this statement repeatedly you demonstrate how little more you have to add.

These are plausible speculation, but your statements are false because you misrepresent Hinton's speculation as established fact.

Hinton both justified and demonstrated that his system is functionally analogous to the noise present in axon potential frequency modulation. If you didn't understand that part of the presentation, then that's simply too bad.

Moreover, claiming that speculation is automatically false is epistemologically gibberish. A speculative statement is neither right nor wrong unless it has been proven or disproven. If you had actually come to me and said "those statements are unproven" or "those statements are speculative," then we would be having a much more civil discussion about this.

Why is it that every time I manage to force an explanation out of you, you make an obvious mistake that proves you don't understand and don't actually know anything about what you're arguing about?

Only people like you, who keep writing scientific nonsense. Unfortunately, nothing seems to stop you.

You've made a mountain out of one vagary, and have tried to question a huge number of reasonable statements without any evidence of actually understanding what they're about. You put no effort into constructively addressing concerns, hurling insults without hesitation and hiding your reasons and justifications as long as possible. If you tried that bullshit at a conference you'd be banned within hours. You're not qualified to be the arbiter of scientific authority here or anywhere else. Stop pretending you are.

Re:Missing the point as usual

[Anonymous Coward]

How do genes know whether they are dominant or recessive?

Re:Missing the point as usual

[Samantha Wright]

"Know" is a bit of a stretch, and "dominant or recessive" is a bit of a high school simplification, but: an allele (a version of a gene) is dominant when it produces an effect that overrides another one. Genes are actually blueprints for proteins, which function as little machines inside of the cell once they're made, so you can see how, for example, if you have a broken version of a gene that doesn't produce a working protein, then the result will usually be recessive. A dominant allele is one that isn't cancelled out by an alternative version. That's why albinism is recessive—the broken copy is hidden behind the working one.

A lot of genes don't display total dominance, though: for them, both copies are important and they contribute more-or-less equally, so a mixture produces an in-between state. How that in-between state manifests can be very complex; it may be a perfect averaging, or an above-normal averaging, or different cells may pick a copy at random (especially in X-linked genes in female animals) that cause splotches or roan.

What's the point?

[Anonymous Coward]

AI researchers nowadays are focused on doing what's possible. That's a good thing; you can't say that much for 1980s AI researchers.
No one's ever seen a human-level intelligence without our astronomical neuron count and human upbringing, and no one can say it's even possible.

Re:What's the point?

[Samantha Wright]

You're thinking of machine learning, which is a separate branch of AI that's more like an overfunded brand of applied statistics—their strategy is actually still to try and push the envelope (like Hinton, another U of T prof, did last year with dropout networks) but they do so in a more results-driven manner. The ML field as a whole is still sore from three or four decades of overpromising on the future, so they try to put their words where their mouths are, and focus on things that are attainable.

Levesque is in the knowledge representation group, which is more closely in step with cognitive science (the leading edge in modelling human thought) but still very philosophical in their approach. KR was the dominant AI field in the 80s (when Prolog and expert systems were all the rage) but it's matured a great deal since then. Here is his homepage, just to show you how different things are now.

Remember that neural networks aren't magic irreducible fairy dust: they're incredibly powerful, but at the end of the day there must be some program that is running within the network unless it's just a wildly complex ever-changing mapping function, which is unlikely given the illusion of consciousness. Given that quantum mechanics is believed to be Turing-complete, it's fairly likely we'll eventually discover some underlying model that lets us produce a human-like cognitive system without the same level of hardware parallelism that the brain has.

Re:What's the point?

[Anonymous Coward]

Was this post done by a marketing agent? It has all the tomfooleries of applied hype terms like:

- Machine Learning
- Knowledge representation
- Neural Networks
- Quantum Mechanics
- Turing complete

Madam, here is my money. I will buy whatever it is you are selling.

Re:What's the point?

[Samantha Wright]

They're real words, I swear. Although we usually just say ML, KR, nets, and QM, if that helps. Here's the thing about QM and Turing completeness. Also, a marketing post wouldn't admit KR was a load of crap in the 80s and ML totally failed to deliver in the 70s.

An eskimo would have the same problem

[giorgist]

An eskimo would have the same problem, does that mean he cannot understand people ?

Re:An eskimo would have the same problem

[The MAZZTer]

The Eskimo would still be able to answer the question ("I don't know.").

Re:An eskimo would have the same problem

[Anonymous Coward]

Or at the very least ask "What's an alligator?"

Re:An eskimo would have the same problem

[Anonymous Coward]

So can some computer programs: Watson includes a confidence percentage in its answer.

Re:An eskimo would have the same problem

[Kjella]

An eskimo would have the same problem, does that mean he cannot understand people ?

In this case he wouldn't understand, but because he lacks knowledge not intelligence. Show him an alligator and a 100 meter hurdles race and he'll be able to answer but the AI will still draw a blank. Ignorance can be cured but we still haven't found a cure for stupid, despite all efforts from education systems worldwide. No wonder we're doing no better with computers.

Re:An eskimo would have the same problem

[Anonymous Coward]

I would think a computer having a biomechanical model for alligators and a spatial ability to recognize hurdles might well be able to conclude that alligators lack the ability to navigate over hurdles, after it runs a bunch of genetic algorithms. Heck, it might even find a solution to getting over hurdles and inform you that alligators can in fact handle hurdles. You might be surprised at the solution it proposes (as they did with crosstalk to improve chip performance).

Re:An eskimo would have the same problem

[Anonymous Coward]

Then ask these question to a software developer and you get answers that no ai can comprehend.

BTW. Alligator could very well do hurdles only victims could not tell as they have been eaten at the end of a run.

Re:An eskimo would have the same problem

[Anonymous Coward]

You've missed the point of the paper and AI. The question is how to make AI answer the question like a human being would since that kind of AI would be very useful for many applications. If you ask AI "can I fly to Sydney for 500 bucks?" you don't want it to start designing an inexpensive jet pack for you...

Re:An eskimo would have the same problem

[profplump]

And a human would only answer the way you wanted them to if they knew that by "fly" you meant "book a seat on a commercial scheduled passenger flight". Which again is about knowledge, not reasoning. Otherwise they (and a similarly intelligent computer) would say no on the same basis as the alligator-hurdles question -- because you are biomechanically incapable of flying to Sydney.

Re:An eskimo would have the same problem

[Anonymous Coward]

Did you read the paper linked to or have you read any other scientific paper on AI? What you refer to as "knowledge" is not knowledge that exists in any such form that it could be readily fed to AI because that kind of "knowledge" is unnecessary to write down because we humans use our reasoning abilities instead. The knowledge in this case is "fly" = take a commercial flight, do what birds do or use a jet pack or other aid to do what birds do. Any big data we could feed AI would give it all three meanings - unintentionally or intentionally, you can decide yourself whether you consider it good or bad considering that we want very capable AI. The AI would then use reasoning to form what you refer to as knowledge so that it can answer the question. The Eskimo would with ease know what answer is being sought when the example question was stated even if he/she simultaneously had the ability to also answer the question "can I build a jet pack for 500 bucks and fly to Sydney?". Meaningfully and e.g. say "not with currently available technology". However, AI that had the data needed to answer both questions would have difficulties and if you read the paper that was linked to you would also understand why it's not a working solution to let AI assume that the answer being sought is the most common one when the question is ambiguous.

Re:An eskimo would have the same problem

[pipatron]

Actually yes, I would like it to design an inexpensive jet pack for me, that would be awesome.

Re:An eskimo would have the same problem

[jkflying]

I was thinking that too. If it could get me to Sydney for $500 by jetpack I'd happily take that option.

Re:An eskimo would have the same problem

[Anonymous Coward]

Showing a picture of an alligator and a picture of hurdles, to an eskimo or native who have never seen such things before, might yield interesting results.
Our knowledge is tightly intervowen with our past, culture, power structures, assumptions and aspirations (sadly: make more money, extract more, these days).

Captcha: imprint

Re:An eskimo would have the same problem

[Anonymous Coward]

If you ask AI "can I fly to Sydney for 500 bucks?" you don't want it to start designing an inexpensive jet pack for you...

Why not? If you had asked the same question of Thomas Edison, he likely would have tried to answer that question with some sort of powered balloon strategy. And, despite what many here feel about his business practices, it's an almost impossible argument to make that Edison wasn't an incredibly intelligent man.

Intelligence deprived of facts/information can seem to be something else entirely.

Re:An eskimo would have the same problem

[Anonymous Coward]

Although you could argue that an alligator could run a 100-meter hurdles race; however, by failing to make any of the hurdles, it would be assessed the same time penalty that a human racer would if they knocked over every hurdle (and thereby failed to make any hurdles at all).

copypasta

[Gothmolly]

For chrissakes, all submitters and 'editors' are doing is copying articles from other sources - nothing original, no EDITORIALIZING, nothing. Slashdot = crappy RSS feed.

Re:copypasta

[FunPika]

You didn't realize this years ago?

Re:copypasta

[newcastlejon]

There's a big difference between editing and editorialising. The former is something I like to see on /. (but seldom do), and the latter is something I never like to see here.

Look up "editorial" and you'll see.

Re:copypasta

[buchner.johannes]

Not just that. The 'article' is not a scientific article, published or accepted in a Journal, but just a blog entry parsed through pdflatex. With sentences like "My feeling is that" it's obvious this won't pass peer review in this form. This seems to be quite popular in Computer 'Science' these days -- you can say you wrote a 'scientific article' without caring about whether its novel or sound, when all you did was to make a brain dump of your half knowledge.

Re:copypasta

[ColdWetDog]

No, it's not a peer reviewed research article. It's a modified lecture given at some conference.

This is a classic venue for opinion pieces / overviews / misogynist rants. Some presumably well known academician gives a 'distinguished talk' and it gets transcribed, cleaned up a bit and placed in a (usually) middling level journal.

I have no idea who this person is, nor his qualifications, nor the status of the journal in question. But it's a well known approach to publishing in various scientific fields and is usually done to get people arguing^Hdiscussing the issues.

Re:copypasta

[Anonymous Coward]

Sigh. This is a written account of a lecture presented as part of Levesque receiving the Research Excellence prize. The first footnote of the paper says so:
"This paper is a written version of the Research Excellence Lecture presented in Beijing at the IJCAI-13 conference. Thanks to Vaishak Belle and Ernie Davis for helpful comments."

Premier conferences don't give these prizes to just anyone, and the opinions of folks like these are worth thinking about.

From the IJCAI website http://ijcai13.org/program/awards (Google cache version, since the original seems to down):
"IJCAI-13 Award for Research Excellence
The Research Excellence award is given to a scientist who has carried out a program of research of consistently high quality yielding several substantial results. Past recipients of this honor are the most illustrious group of scientists from the field of Artificial Intelligence;
They are: John McCarthy (1985), Allen Newell (1989), Marvin Minsky (1991), Raymond Reiter (1993), Herbert Simon (1995), Aravind Joshi (1997), Judea Pearl (1999), Donald Michie (2001), Nils Nilsson (2003), Geoffrey E. Hinton (2005), Alan Bundy (2007), Victor Lesser (2009) and Robert Anthony Kowalski (2011).

"The winner of the 2013 Award for Research Excellence is Hector Levesque, Professor of Computer Science at the Department of Computer Science of the University of Toronto. Professor Levesque is recognized for his work on a variety of topics in knowledge representation and reasoning, including cognitive robotics, theories of belief, and tractable reasoning."

Re:copypasta

[Anonymous Coward]

Actually, IJCAI is the top conference in the field of Artificial Intelligence and every published paper goes through extensive peer review.

Computer Science is a bit different from most other science in that top conference proceedings (IJCAI, NIPS, ICCV, CVPR, etc.) have the weight of a journal. In fact, publishing there is more prestigious than most journals. Review period lasts 3-4 months and includes a rebuttal phase, like a journal.

This paper looks like an invited lecture or a position paper expected to provoke a debate, that is true. But calling IJCAI "some conference" is like calling Nature "some newspaper".

Re:copypasta

[smittyoneeach]

What's the opposite of artificial intelligence? "Natural ignorance."

*People* can't understand people

[msobkow]

People are irrational. They ask stupid questions that make no sense. They use slang that confuses the communication. They have horrible grammar and spelling. And overseeing it all is a language fraught with multiple meanings for words depending on the context, which may well include sentences and paragraphs leading up to the sentence being analyzed.

Is it any surprise that computers can't "understand" what we mean, given the minefield of language?

Re:*People* can't understand people

[rolfwind]

Well, at least I'm not the only one who appreciates a computer's relative unambiguity (aside the ones programmed into it).

Re:*People* can't understand people

[retchdog]

Great, so once we are all speaking lojban, AI will be a piece of cake, right?

Re:*People* can't understand people

[buchner.johannes]

Just ask the question in Lojban?

Re:*People* can't understand people

[rabtech]

People are irrational. They ask stupid questions that make no sense. They use slang that confuses the communication. They have horrible grammar and spelling. And overseeing it all is a language fraught with multiple meanings for words depending on the context, which may well include sentences and paragraphs leading up to the sentence being analyzed.

Is it any surprise that computers can't "understand" what we mean, given the minefield of language?

Even if you came up with a regular easy to parse grammar it wouldn't help. Even if you fed the computer all known information about alligators and hurdles in that standard format it wouldn't help. That's the point... Now that we are starting to do much better at parsing and reproducing speech, it turns out that isn't really the hard problem.

Re:*People* can't understand people

[msobkow]

That's the whole point about "context", though. It's not just the context of the sentence at issue, but the context of the knowledge to be evaluated, the "memory" of the computer if you will. It's an exponential data store that's required, and then some, even when using pattern matching and analysis to identify relevant "thoughts" of memory.

Re:*People* can't understand people

[fuzzyfuzzyfungus]

"Is it any surprise that computers can't "understand" what we mean, given the minefield of language?"

It is certainly no surprise that computers can't; but since we know that humans can (to a substantially greater degree), we can say that this is because computers are far worse than humans at natural language, not because natural language is inherently unknowable.

Re:*People* can't understand people

[fuzzyfuzzyfungus]

Great, so once we are all speaking lojban, AI will be a piece of cake, right?

Only if we are speaking lojban on the semantic web. And after we've abandoned empiricism for syllogism.

Re:*People* can't understand people

[antdude]

I don't understand you. [grin] :P

Re:*People* can't understand people

[colinrichardday]

How do we deal with multiple quantification by syllogism?

Re:*People* can't understand people

[colinrichardday]

Is it any surprise that computers can't "understand" what we mean, given the minefield of language?

The problem isn't entirely linguistic. Humans can communicate because we have an awareness of a common reality. Until/Unless computers are also aware, they will have problems understanding us.

Re:*People* can't understand people

[AthanasiusKircher]

The summary is problematic. The alligator example is interesting, but the later examples in the article are better. Most of them don't depend on "imagination" or "creativity" or whatever to answer the question, or on a large bank of cultural knowledge, but only a basic knowledge of what words mean in relationship to each other. Yet AI would often fail these tests.

People are irrational. They ask stupid questions that make no sense.

While this is true, it has little bearing on the issues raised in TFA. It's also unclear what you mean by things that "make no sense." If you mean that literally, as in mentally challenged people babbling nonsense, then I do not expect a computer to be able to answer nonsense anymore than a normal person could. If 99% of adult native English-speakers without severe mental problems can answer a simple question correctly, I expect a computer that is said to "understand English" to be able to do the same.

If you mean -- as many geeks do when complaining about language imprecision -- that people ask questions without the precision used in formal made-up languages (like programming languages or stereotyped logic statements), well that's a hopelessly incomplete view of what "meaning" is. We use natural language despite its seeming imprecision because it actually can convey incredibly complex webs of meanings rather efficiently, instead of only allowing a specified small set of particular relationships that formal "rational" languages can use to produce a very limited set of meanings.

"Irrationality" and "making no sense" don't matter if 99% of native speakers can answer a simple question without hesitation. That means the the question seems both perfectly "rational" and "makes sense" to English speakers, and the same should be required of any computer said to do the same thing.

They use slang that confuses the communication. They have horrible grammar and spelling.

Again, a separate problem that's not very relevant to the concerns in TFA. As we've seen with improvements in Google search corrections, autocorrect technologies, etc., these issues are probably relatively minor to deal with compared to understanding the underlying meaning of standard natural language.

And overseeing it all is a language fraught with multiple meanings for words depending on the context, which may well include sentences and paragraphs leading up to the sentence being analyzed.

The examples given in TFA are things like simple 2-3 sentence scenarios where all the required information is contained in those sentences. The answer required is often a simple multiple-choice.

For example: "Joan made sure to thank Susan for all the help she had given. Who had given the help? a) Joan b) Susan"

Yes, you're talking about a much larger issue of context, but the examples in TFA pinpoint much smaller-scale failures to comprehend natural language. Many of the questions depend on simple patterns where 3 or 4 words used together in a sentence establish particular relationships among those words that any native speaker would get. Being able to parse those connections is what it actually would take to understand what those 3 or 4 words "mean."

Meaning is not atomic, and it is not only based in single words (which is your point). It exists in everything from phonemes and parts of words like prefixes, roots, and suffixes (that establish potential associations from sounds and grammatical clues about how the word functions) through phrases, sentences, and entire paragraphs.

But this is not a failure of language. It is how language fundamentally works. Words don't really have "multiple meanings": they only come to mean anything when connected with other words. We only have the illusion that individual words have specified meanings because dictionaries have been constructed along that model. It's a useful way to think about meaning, but it has little t

Re:*People* can't understand people

[msobkow]

I quoted "understand" because there are many levels of understanding from mapping the atomic meaning of words based on sentence structure through to contextual awareness and full scale sentience.

Re:*People* can't understand people

[Samantha Wright]

No, we'd go mad because the spelling system is a trainwreck of unparalleled proportions.

Re:*People* can't understand people

[Tablizer]

People are irrational.

You may be confusing "social games" with "irrational". it's possible playing with words is part of a verbal social dance, not unlike peacock displays. Are peacocks "irrational" for doing such?

A Vulcan view of reality may not match up well with the realities and "oddities" of natural selection.

Re:*People* can't understand people

[martin-boundary]

Physicists are irrational. They ask stupid questions that make no sense to anyone else. They use jargon that confuses the communication. They have horrible grammar and spelling. And overseeing it all is a physics language fraught with multiple meanings and everyday words used in highly technical ways.

Any sufficiently advanced group of people are indistinguishable from an irrational group to anyone else, human OR robot.

To put the above in technical language: knowing the rules of propositional calculus alone is useless without knowing the axioms, for proving theorems.

Re:*People* can't understand people

[93 Escort Wagon]

On the old TV series "Get Smart", there was a robot agent named Hymie. There was a running gag where one of the human agents would say something along the lines of "kill the lights, Hymie", whereupon the robot would of course pull out his gun and shoot the light bulbs; or tell Hymie to "hop to it" (which led Hymie to start hopping), "get the door" (so he'd rip the door off its hinges and bring it over to Max), etc.

Re:*People* can't understand people

[retchdog]

well, once we are all insane, it would be easier to find an acceptable AI. unfortunately, it would also be harder to develop. :-/

Re:*People* can't understand people

[profplump]

"Natural language" is only "natural" for the particular organisms who have developed the ability to use it. Computers have no trouble understanding their own "natural" languages.

Re:*People* can't understand people

[profplump]

Exactly which languages are not "made up"?

Re:*People* can't understand people

[Anonymous Coward]

Shit like this is exactly what he was talking about in regards to geeks. You know what he meant.

Re:*People* can't understand people

[3seas]

"Is it any surprise that computers can't "understand" what we mean, given the minefield of language?"

Uh, computers don't "understand" anything, they are machines that simply do what they are programmed to do.
The first step is for humans to understand what computers really are. They are nothing more than abstraction processing machines which have not the ability to "understand" the abstractions they process but only to process abstraction as they are programmed to do.

Artificial Intelligence is artificial by definition. And the appearance of intelligence in computers is nothing more than an active image of human thought processes captured and put into the stone of computer hardware. So to increase the "appearance" of intelligence we only need to capture more human thought processes and map them in a manner that is searchable.

Of course the way to do this is to recognize the functions we humans cannot avoid the use of and program the computer to have this functionality, that we may be better able to capture and map images of human mental processing in a manner of machine processing ability.

When the software industry finally lets go of their hold on teh users and let the users do more for themselves, we will reach this "Appearance of intelligence" in machine much faster. See: http://abstractionphysics.net/pmwiki/index.php .

Re:*People* can't understand people

[fuzzyfuzzyfungus]

I was attempting a joke. Syllogisms have friends because they are relatively easy and provide easy access to valid (albeit not necessarily true) conclusions; but aren't actually of any significant use for attacking problems in the world.

Simple

[Anonymous Coward]

People don't understand people.

Helps to remember...

[djupedal]

There are two basic forms. One involves training the human on the commands the computer will respond to properly and the other involves training the computer to recognize an individuals speech patterns.

IBM has been busy for some time working on real-time translators, and I think that path is where the future lies, not just in a voice command TV.

Re:Helps to remember...

[ultranova]

There are two basic forms. One involves training the human on the commands the computer will respond to properly and the other involves training the computer to recognize an individuals speech patterns.

And neither helps here. The fact is, you don't know if an alligator can run the hundred-metre hurdles. When you're asked to answer the question, you imagine the scenario - construct and run a simulation - and answer the question based on the results. In other words, an AI needs imagination to answer questions like these. Or to plan its actions, for that matter.

Re:Helps to remember...

[houghi]

Indeed I do not know if it can run the hurdles. I do not know the rules. Are you required to jump the hurdles, or can you run under them or even through them by pushing them over? If so, it would be possible, because the fact that they can not jump them becomes irrelevant.

As seen here, we see already different answers. to one question. These vary from yes to no. If the answer is yes, does that mean that the people who answer no are not human?

And just like the computer, I answered that I did not know. Am I a computer?

Where it would become interesting is if you start asking joke questions. e.g.
Q: It is green and if it falls out of a tree, you are dead. What is it?
A: A pool table

Q: What is the difference between a parakeet?
A: Both legs are the same length, especially the right one.

And humor is not even easy for humans to understand. What I learned when learning a new language is that there are several levels. (Very generic.
1) Cursing
2) Ordering things. Telling where you are from and where you go
3) Work related conversation or subjects that you are familiar with
4) Reading the newspaper (As it is written for a majority of people)
5) Advanced discussion on any subject
6) Understanding the jokes (Does not mean you must think they are funny.)

I would say computers are now at level 2 and the question was a level 5 question. The step between 2 and 3 is not that small. step 2 is repeating things. Step 3 is also listening and responding as well as experience in life. Something that computers lack, so they try to go to step 4 directly.

Re:Helps to remember...

[profplump]

"Run a simulation" is typically not understood to be substantially similar to "imagination". Computers are perfectly capable of running all sorts of simulations. They're even fairly adept at constructing models given unorganized input constraints. And I would fully expect either or a human or computer with biomechanical knowledge of alligators and physical knowledge of hurdles and some existing model of physics at an appropriate scale to be able to construct and run an appropriate model to answer the question, with no use of imagination at all.

In this particular example the real difference is most humans would blindly guess, with no simulation at all. They'd either guess that alligators can jump, or that they can't, and answer based on that essentially unsupported guess. Only someone who had significant knowledge of alligator movements would give even a vaguely reliably answer to such a question. Which is exactly the same state computers are in.

Re:Helps to remember...

[ultranova]

"Run a simulation" is typically not understood to be substantially similar to "imagination".

But of course it is. Constructing a model and evolving it based on some ruleset is the very definition of simulation. Humans simply like to represent the input and results in terms of sensory stimulus rather than numbers.

In this particular example the real difference is most humans would blindly guess, with no simulation at all. They'd either guess that alligators can jump, or that they can't, and answer based on that essentially unsupported guess.

Most people have at least a vague idea of what alligators are and how they move and would base their answer on that. And in fact, when someone says "alligator", what comes to your mind? An animated image of a crawling beast. A simulator. Those are what human brain stores and recalls, not factoids like whether alligators can jump hurdles. And why not? If the abilities of alligators is ever going to have some impact on you beyond academic curiosity, it's likely you need the relevant info in a hurry, and simply reactivating a stored simulator is faster than building one from known facts.

Which, on an unrelated topic, is why humans have such problems remembering passwords, but it becomes trivial if you can attach a story - a simulator - to one which generates the passcode. And also why witnesses tend to be unreliable: a remembered past event is stored as a simulator, which (re)generates sensory stimuli associated with the event, and will happily fill gaps with procedurally generated garbage which the witness then represents as a sworn testimony - not because he's a liar, but because remembering is the same as imagining for humans.

All of which means that strong AI, should it ever exist, will likely be just as prone to irrationality as humans are.

Re:Helps to remember...

[Anonymous Coward]

I don't see why that imagination should be a problem. Computers are already able to imagine problems far more complex that humans...engineers now use CAD programs to design and test out their ideas in a way that wasn't possible previously.

What would be necessary would be to model an alligator in a CAD-like environment that understands they physics involved. You wouldn't need to have seen an alligator move previously if you can look at the length of the legs and the weight of the animal and realize the force that those legs would need to generate to jump over a hurdle given the physics restrictions placed by our planet.

It's the ability to map abstract concepts or images onto the models that computers excel at that needs to happen.

I disagree

[Anonymous Coward]

Alligators certainly can run the 100 meter hurdles. They just can't do it well.

Re: I disagree

[Anonymous Coward]

hell, i could run the hundred meter hurdles. especially if a gator is chaising me.

Computers understands humans

[CmdrEdem]

Through a thing called programming language. The same way we all need to learn how to speak with one another, we need to learn how to properly communicate with a computer.

Not saying we should not teach machines how to understand "natural" language, text interpretation and so on, but that headline is horrible.

Well, duh

[Anonymous Coward]

Do we need an article in the New York Times to make us realize that a machine put together with today's technology doesn't come close to having what humans traditionally regard as intelligence, regardless of how well it does on someone's lame attempt at a Turing Test? Only an arm-waving, James Spader-lookalike professor or startup founder would argue otherwise.

computers and people

[Skapare]

Computers cannot understand anything. Nothing can understand people. And someone expected computers to somehow understand people? Now that's a corner case for ya.

Re: computers and people

[Anonymous Coward]

If computers can understand nothing, and nothing can understand people, then by the transitive property computers can understand people.
Q.E.D. ;)

Re:computers and people

[RespekMyAthorati]

And no one can understand you!

Does AI have to understand all aspects of people?

[Anonymous Coward]

Don't make me wait for automatic cars because the AI can't answer the question "Can an alligator hurdle?"

Also FTFA:

As a field, I believe that we tend to suffer from what might be called serial silver bulletism, defined as follows:
the tendency to believe in a silver bullet for AI, coupled with the belief that previous beliefs about silver bullets were hopelessly naive.
We see this in the fads and fashions of AI research over the years: first, automated theorem proving is going to solve it all; then, the methods appear too weak, and we favour expert systems; then the programs are not situated enough, and we move to behaviour-based robotics; then we come to believe that learning from big data is the answer; and on it goes.

Spelling and grammar errors

[roman_mir]

While this was not deliberate (FTFA):

The large ball crashed right through the table because it was made of Styrofoam. What was made of Styrofoam? (The alternative formulation replaces StryRofoam with steel.)

a) The large ball
b) The table

I imagine using bad spelling would not deter computers today (Google checks spelling and offers auto-corrections), but understanding bad grammar is a harder problem to solve for computers, so never mind convoluted questions, how about just using bad grammar to figure out who the real people are vs. who the computers are?

Wait a minute, is that what all the bad grammar (and spelling and punctuation) on the Internet all about? All this time I thought people just can't write well most of the time and actually what it may be is a way to distinguish between the real people and robots.

Re:Spelling and grammar errors

[Tablizer]

Rumor has it Mythbusters killed a puppy with a Styrofoam cannonball.

Re:Spelling and grammar errors

[Zontar The Mindless]

That's an issue with semantics, not grammar.

Re:Spelling and grammar errors

[Zontar The Mindless]

Guess again, sockpuppet.

Hook, line, and sinker.

Do better.

[FatLittleMonkey]

http://slashdot.org/submit

people can only answer questions they know

[alen]

the other day my almost 6 year old said we live on 72nd doctor. the correct designation is 72nd Dr
since doctors use dr as shorthand, he thought streets use the same style

Re:people can only answer questions they know

[Anonymous Coward]

Bright kid...give him a cookie and explain the alternate meanings for the abbreviation Dr.

Re:people can only answer questions they know

[Anonymous Coward]

So... Who is the 72nd Doctor?

Fans everywhere want to know!

Re:people can only answer questions they know

[Tablizer]

This dude.

Re:people can only answer questions they know

[Anonymous Coward]

In other words you are saying that your offspring does not know where he lives.

Re:people can only answer questions they know

[Anonymous Coward]

I'm afraid you have to wait for all that wibbly-wobbly stuff to pan out first. Otherwise - spoilers...

Missing human "imagination"

[Brian_Ellenberger]

The thing missing with many of the current AI techniques is they lack human "imagination" or the ability to simulate complex situations in your mind. Understanding goes beyond mere language. Statistical models and second-order logic just can't match a quick simulation. When a person thinks about "Could a crocodile run a steeplechase?" they don't put a bunch of logical statements together. They very quickly picture a crocodile and a steeplechase in a mental simulation based on prior experience. From this picture, a person can quickly visualize what that would look like (very silly). Same with "Should baseball players be allowed to glue small wings onto their caps?". You visualize this, realize how silly it sounds, and dismiss it. People can even run the simulation in their heads as to what would happen (people would laugh, they would be fragile and fall off, etc).

Re:Missing human "imagination"

[Tablizer]

But it takes experience to determine such a simulation would be "silly".

A compromise may be to have the computer consider all possible interpretations and then probe the user for specifics.

For example, "Do you mean the actual animal, cartoons and fictional animals, or the Florida Gators sports team?"

Human: "I mean the actual animal"

Computer: "I could not find any instance of actual alligators jumping hurdles."

Or maybe, "Here is an image of an alligator jumping hurdles, based on the caption. I cannot confirm it is authentic."

Re:Missing human "imagination"

[profplump]

Exactly how does a human "imagine" the right answer to your steeplechase question? Does the average person have information about how fast alligators can run over long distances? They might know that alligators are quite fast over short distances, but very few people know enough about alligators to accurately "visualize" what an alligator would look like in long-distance overland travel -- I'm pretty sure most humans would just blindly guess.

At best humans might reason "alligators spend most of their time in and around water, so they're probably not well-suited for long-distance running", but that's not the result of any imagination or simulation or visualization -- it's just a basic application of observations and/or learned rules, and computers can do both those things given the same information.

Now, it's fair to say that most computers, including AI systems, wouldn't apply those techniques, but that's a design choice (as TFA points out) not a limitation in the way computers function.

Re:Missing human "imagination"

[Anonymous Coward]

I think you missed the point here.

The problem with aligator running hurdles/steeplechase is not speed, but jumping.

And given the short legs and big heavy body and tail, I can easily imagine the alligator will not be too good at that...

Because they don't understand purpose or intention

[divisionbyzero]

That's why. They don't have desires, fears, or joys. Without those it's impossible to understand, in any meaningful sense, human beings. That's not to say that they can't have them but it's likely to come with trade-offs that are unappealing. And for good measure, they also don't understand novelty and cannot for the most part improvise. All of which are considered hallmarks of human intelligence.

Oblig. XKCD

[FatLittleMonkey]

Perhaps we need a new form of Turing Test where the AI must turn a weird novel query (like "can alligators run the 100m hurdles?") into something Google can understand, and then work out which of the returned sites has the information, parse the info and return it as a simple explanatory answer.

Turing test

[Dan East]

Intelligence implies usefulness. Intelligence is a tool used by animals to accomplish something - things like finding food, reproducing, or just simply staying alive. We've abstracted that to a huge degree in our society where people can now engage in developing and expending intelligence on essentially worthless endeavors simply because the "staying alive" part is pretty well a given. But when it comes down to it, the type of strong AI we need is a useful kind of intelligence.

The problem with the Turing Test is it explicitly excludes any usefulness in what it deems to be an intelligent behavior. From Wilipedia: "The test does not check the ability to give the correct answer to questions; it checks how closely the answer resembles typical human answers." That bar is set far, far too low, and is even specific to a generic conversational intelligence instead of something useful. The Turing Test is far too overrated and synonymous with the field of AI and really needs to just go away. It reeks of the Mechanical Turk kind of facade versus any real metric.

Re:Turing test

[Anonymous Coward]

Well, the problem lies when you have a scientist who's already made up their mind what the conclusion should be, before they've gotten around to the science bit. That tends to lead to attempts to skew the result this way or that to confirm their already existing belief.

AI is somewhat of an interesting idea, but people preaching the AI gospel kind of come across as idealists more than anything. It's good to see someone throw some healthy skepticism into the mix. Even if one side is right, there should never be lack of skepticism, because that is not good science.

Re:Turing test

[Anonymous Coward]

The problem with the Turing Test is it explicitly excludes any usefulness in what it deems to be an intelligent behavior. From Wilipedia: "The test does not check the ability to give the correct answer to questions; it checks how closely the answer resembles typical human answers." That bar is set far, far too low,

No, I don't think that bar is set "too low". You can say whatever you like in a Turing test - so you can of course ask the alligator question. Or any other specific question that need some (common) intelligence for responding properly.

A human might find the alligator question silly or funny, but insist and they will be able to answer.

AI is still hopeless on turing tests. They may be able to answer a single question in a believable way - if the question isn't too contrived. And they may be able to "change the subject" when they don't get the question. But AI cannot hold an entire conversation! Try chatting with an AI, and refer back to something said 5 sentences ago. There is no memory of recent communication. The AIs answers each sentence as if you're meeting for the first time. A Turing test can be many sentences and cover some topic at length - that is when AI betrays itself easily.

Re:Turing test

[Hentes]

It may be too low, but it's still high enough that no AI has ever come close to passing it.

Chris McKinstry's MIST covered this years ago

[blue trane]

http://en.wikipedia.org/wiki/Minimum_Intelligent_Signal_Test

McKinstry gathered approximately 80,000 propositions that could be answered yes or no, e.g.:

Is Earth a planet?
Was Abraham Lincoln once President of the United States?
Is the sun bigger than my foot?
Do people sometimes lie?
He called these propositions Mindpixels.

These questions test both specific knowledge of aspects of culture, and basic facts about the meaning of various words and concepts.

Re:Chris McKinstry's MIST covered this years ago

[SoftwareArtist]

Except that many of these questions can be answered by pure data mining. Is Earth a planet? You'll find a high number of references to "Earth" and "planet" close together, so that suggests the answer is yes. Ditto with "Abraham Lincoln" and "President of the United States". In fact, you can do a trivial grammatical transformation to make the question into a statement ("Earth is a planet"), and you'll probably find many occurrences of that sentence on the web. Contrast that with the questions described in the article, which are carefully designed to make that sort of analysis useless.

Re:Chris McKinstry's MIST covered this years ago

[blue trane]

The wiki article may not have captured McKinstry's full purpose, which was to ask questions of the type the article refers to, which any human knows the answer to, but computers may not have seen before. So the http://aiki-hal.wikispaces.com/file/detail/gac80k-06-july-2005.html (list of questions assembled by Chris) includes such questions as:

Is a car bigger than a breadbox?
Are horses bigger that elves?
Is an elephant bigger than a cat?

etc.

These sentences, transformed into declarative form, have probably not occurred on the web, which was the point of McKinstry's test.

Consider also the misspellings and grammatical mistakes in the questions, which humans are nonetheless able to answer, but which are unlikely to have been part of any web-gathered corpus...

AI = Artificial Imbecile

[Anonymous Coward]

The state of AI may have reached the level of Imbecile in some labs but in the real world its still stuck at Idiot. I know this is not exactly the best example but lets examine the phone AI software that many companies think is a good idea to be what answers when a customer calls for service. From my experence it goes like this.. I'm on the road, no internet available so I use my cell to pay a bill, no I don't have an app for that.. -AI- "hello welcome to XYZ" please say what you are calling abour or you could say "new customer" or "balance due". -me' "I want to make a payment" -AI- "Ok I understand you would like to make a payment, please say yes, no or other" -me- "yes" -AI- "I'm sorry I didn't get that, please say yes, no or other". -me- "what ?" -AI- "Ok great, one moment while I transfer you to open a new account". -me' "No I want to make a payment". -AI- "I'm sorry I didn't get that, to proceed with the new account please tell me what credit card or bank account you would want to use" -me- you'er a fucken Imbecile.. did you get that ? "

Re: AI = Artificial Imbecile

[Anonymous Coward]

Just swear Angrily. More than a few of these things have an unadvertised shortcut to a live person when the customer seems pissed off.

A bit of hyperbole...

[blahplusplus]

... Human beings are not THAT radically different from computers in that our electric circuits ARE comparable to some extent with electronics because they both use matter and energy and possess similar natural electrical characteristics. Although they may be shaped differently. I'm certain there are MANY insights and cross over ideas and concepts from computer science and how it applies to mind, and the reverse, how biological concepts apply to computer science and electrical circuits in general.

Let's not forget both the brain and computers are made of atoms, electrons, etc. So there's going to be some cross over whether people like it or not.

We can think of the brain as a big electric converter made of biological molecules that converts electromagnetic signals from one form to another (in a general sense) it just does it in foreign way which we think is 'radically different', but more or less the mind is just a DIFFERENT take on a fundamental theme (fundamental natural relationships) of how nature can organize and process information. It's just that no one has sat down to find and expose all the relationships and expose them in detail. So that the unlearned and those who are invested in romantic view of mankind can be disqualified from the conversation.

The Trouble with Turing

[Capt.Albatross]

The problem with most proposed tests for intelligent computing is that not everything that humans need intelligence to perform require intelligence. For example, Gary Kasparov had to use his intelligence to play chess essentially with the same performance as Deep Blue, but nobody, especially not its developers, mistook Deep Blue for an intelligent agent.

A recent post concerned AI performance on IQ tests. The program being tested performed on average like a 4 year old, but, significantly, its performance was very uneven, and it did particularly poorly on comprehension.

Turing cannot be faulted for not anticipating the way Turing tests have been gamed. I think his basic approach is still valid; it just needs to be tweaked a bit. This is not moving the goalposts, it is a valid adjustment for what we have learned.

Re:The Trouble with Turing

[SoftwareArtist]

It's important to distinguish between weak and strong AI. When a human plays chess, we consider that to be an act of intelligence, even without having any idea what's going on in their brain. We therefore need to accept a computer that plays chess as also being intelligent. Ditto for translating a document from German to English, or figuring out the best route for driving to the airport. When a human does these things, we call it intelligence. Our judgement that they are "displaying intelligence" is not based on understanding how they did it. We therefore must accept a computer that does them as being intelligent too.

But this is weak AI. It can do the specific tasks it has been designed to do, and may do them extremely well. But it isn't general. If you give it a new task it wasn't designed to do, it can't analyze the task and figure out how to do it. That would be a strong AI, and that's what we haven't managed to create.

Re:The Trouble with Turing

[RespekMyAthorati]

Turing never suggested that any particular test could establish that a machine possesses human-like intelligence.

If a machine fails a Turing-type test (because people can easily tell that it is a machine) then the current design can be discarded.
But, if the machine passes the test, that only means that the system can then go on to try another, harder test.

Where does this end? Maybe never, since there may always exist a future test that it will fail.
However, if a system goes, say, ten years without failing, a lot of people will probably be satisfied with that.

Re:The Trouble with Turing

[Capt.Albatross]

When a human does these things [play chess, translate documents, plan a route], we call it intelligence. Our judgement that they are "displaying intelligence" is not based on understanding how they did it. We therefore must accept a computer that does them as being intelligent too.

The fact that we cannot explain intelligence does add some difficulty to talking about it, and while reasonable people can disagree over this, I think the approach you use here tends to confuse two things, which you then have to separate by introducing the qualifiers 'weak' and 'strong'.

In your argument, you consider intelligence to be an attribute of an activity, such as playing chess, and then you transfer this attribute to any agent performing the activity. But how do you decide which activities are intelligent? My guess is that you do it by the human test: if humans use intelligence to perform the act, then it is an intelligent act.

There is a sort of semantic flow here, and it has formed a self-referential circle: intelligent actions bestow the quality of intelligence on the agents performing them, but the actions were bestowed the quality of intelligence in the first place by the agents performing them.

The other problem with this approach is that you still have to make a completely independent judgement of what constitutes intelligent action: if moving from A to B or eating were to be included, all animals are intelligent, and the term loses its usefulness. If you can make the distinction for a human's action, you can do it for any agent's actions without invoking semantic flow. By this analysis, we can see that intelligence is a property of the means by which things are done, not the actions themselves.

So the semantic flow approach has not helped in the definition of intelligence, but it also adds another difficulty. Out of the specific examples you give, a computer finding the best route too the airport is not considered to be AI, weak or otherwise. In the case of Deep Blue, even its developers were at pains to say it was not AI, because we know that the techniques it used, such as exhaustive search to considerable depth and a huge memory of pre-computed end-games, were not options for Kasparov, yet he played essentially as well as the computer. Whatever the intelligence he used was, it was not what the computer was doing.

So the problem is that the semantic flow approach to defining AI would label both Deep Blue and my car's GPS as intelligent, and arguably also the car itself for merely moving from A to B (at least when the automatic parallel parking feature is in use.) To get around this problem, the term 'weak AI' was introduced, with the distinction that 'weak AI' is not generalizable.

If you take my approach, and consider intelligence to be an attribute of the way in which something is done, rather than of either the actor or the action, then you get more directly to a similar position, but without the confusion you get whenever the strong/weak distinction is not made, and one party says AI is a solved problem, and the other says that's nonsense.

Furthermore, the strong/weak workaround is a bit of a kludge that may be OK for these examples, but it runs into difficulties with machine learning, which is generalizable to a degree. I think it is still an open question whether machine learning has, or ever will, achieve intelligence. Consider your third example, translation. Machine learning does it (and impressively so) by a thorough (if not exhaustive) calculation over correlations in a very large sample set. Humans, however, approach it differently: we go through the intermediate step of understanding the text.

Understanding is a concept that we cannot explain, but that does not render it useless, because we understand it well enough to recognize it. The study of an AI's performance on an IQ test is relevant here, because the AI did particularly badly on questions that required understanding (the 'why' questions, as the article p

Re:The Trouble with Turing

[Capt.Albatross]

Where does this end? Maybe never, since there may always exist a future test that it will fail.
However, if a system goes, say, ten years without failing, a lot of people will probably be satisfied with that.

I think that, in the situation you predict, we will find it more useful to recognize degrees of intelligence, rather than talk in pass/fail terms.

Re:The Trouble with Turing

[SoftwareArtist]

It sounds like you want to define intelligence entirely in terms of how you do something, not in terms of what you do. But that raises a difficult question: what techniques count as "intelligence"? And how do you decide that?

Consider playing chess. You say that computer chess programs are not intelligent because we understand how they work, and those are different from what a human does. But human chess playing is actually pretty easy to understand:

1. Inexperienced players work almost entirely by trying to predict the next few moves: "If I do this, what will my opponent do next?" They aren't very good at it, because they aren't very experienced.

2. Slightly more experienced players still do this, but they're better at it, and they also have added some heuristics to decide what is a good outcome, like, "I want to control the center of the board." They have memorized the standard point values attached to pieces, so they can better answer questions like, "Is trading a rook for two pawns a good idea?" And they have memorized some standard opening sequences that they can use without having to really think at all.

3. Expert players continue to do all of these things, but much better. They also have added a third approach: they have a fantastic memory for board configurations they have seen in the past, and can recognize at a glance that a particular arrangement of pieces is one they know, and that it leads to a win for black.

That is what humans do. And all of these techniques are ones that computer chess programs can also be programmed to use. You want to say that human chess players use some mysterious power called "intelligence" that is different from what computers do. But that just isn't true.

Likewise, you say that humans translate a text by "understanding" it, which you define with the Potter Stewart approach of, "We can't explain it, but we know it when we see it." But do we really know it when we see it? If we can't explain it, how can we be certain we actually are seeing it when we think we are? Neuroscience still has a long way to go in fully explaining what happens in your brain when you "understand" something, but based on what we know so far, it is entirely possible that human understanding is nothing more than symbol manipulation. Your brain contains lots of symbols (each stored as a pattern of neuron activation) that represent objects, concepts, etc. And you have learned relationships between those symbols, and rules for manipulating them. Exactly like a computer does.

Perhaps that isn't really how humans brains work, but based on our current knowledge, it's at least entirely possible. That also, of course, is exactly how computers work. So maybe human understanding is something fundamentally different from any current AI. But it's also entirely possible that it isn't. Until we can fully explain human understanding, we have to assume that anything that looks like understanding may actually be it.

And even if we determine that human understanding is fundamentally different from AI, that still leaves us with the question of how to define "intelligence". If two systems solve the same problem and reach the same result, what qualifies one as being "intelligent" and the other not? This is ultimately just about how to define a word. We may want to define it in a way that only includes humans and excludes everything else. But do we have any justification for that? Or is it just human prejudice, wanting to be able to claim that we are somehow special?

Re:The Trouble with Turing

[Capt.Albatross]

It sounds like you want to define intelligence entirely in terms of how you do something, not in terms of what you do. But that raises a difficult question: what techniques count as "intelligence"? And how do you decide that?

It doesn't raise a difficult question - the question is difficult no matter how you look at it. Unless you are proposing that an intelligent action is defined as anything that a human does, which has a host of problems that I touched on in my previous post, then, in your approach, we face the same question when deciding which actions are intelligent.

Consider playing chess. You say that computer chess programs are not intelligent because we understand how they work, and those are different from what a human does.

Not exactly - It is not my position that AI has to work like human intelligence. The reason no extant chess program counts as intelligent is because there is clearly more to intelligent chess-playing than brute-force search, a vast memory of pre-computed partial results, and custom heuristics that are inscrutable to the agent. We can see that this is true because a) humans cannot perform the searches or memorize end-games to a degree even remotely close to what the machine is capable of, yet the best human players perform comparably to the best computers; b) if you take those capabilities from the machine, its performance crashes; and c) humans not only understand the heuristics they use, they developed them themselves.

Developing those heuristics is part of intelligent play. I believe some of Deep Blue's heuristics came from machine learning applied to a vast set of games against other programs, which is where things begin to get interesting, but it is not as if Deep Blue figured out what training it needed to do to improve itself.

You don't have to take my word for it - The developers of Deep Blue carefully and deliberately avoided describing it as intelligent.

But human chess playing is actually pretty easy to understand...

That's what an earlier generation of AI developers thought, but after a promising start, they ran into a wall of diminishing returns. It was an early example of the AI community prematurely anticipating success.

Furthermore, you glossed over the difficult bits, such as how the heuristics are developed in the first place.

And all of these techniques are ones that computer chess programs can also be programmed to use.

And yet the humans kept on winning. Deep Blue finally scraped a Pyrrhic win by brute force. The difficulty of that achievement belies your claim that chess is just a matter of a few easily-understood algorithms coupled with memorization.

You want to say that human chess players use some mysterious power called "intelligence" that is different from what computers do. But that just isn't true.

To accurately state what I think, 'mysterious' should be read as 'currently poorly understood' and 'different from what computers do' should be read as 'different from what computing has done so far.' And that I believe to be empirically true, for all the reasons I have given here and in my first post.

On the other hand, you are now arguing against the role of intelligence in chess-playing, which is the opposite of your position in your first post.

Likewise, you say that humans translate a text by "understanding" it, which you define with the Potter Stewart approach of, "We can't explain it, but we know it when we see it."

That is the nature of things that are not well understood, and you can't change that by rephrasing the problem.

But do we really know it when we see it?

You assumed you could when you chose to define an intelligent action as something that an intelligent agent does - specifically, when you decide whether an agent is intellige

Re:The Trouble with Turing

[SoftwareArtist]

I think we're really just arguing about the definition of "intelligence". Of course, there's no "right" answer to questions like that, since all definitions are ultimately arbitrary. We can define a series of syllables to mean whatever we want. But some some definitions are clearly unreasonable, such as if they contradict themselves or conflict with the way everyone else in the world uses the word. So let me explain the requirements I think a reasonable definition of intelligence needs to meet.

Consider the statement, "Humans have intelligence." That's completely uncontroversial. Any neuroscientist would agree with it. So would any person on the street who knows nothing about neuroscience. It was equally uncontroversial 30 years ago when much less was known about the human brain, or 300 years ago when nothing was known about it.

Conclusion: any reasonable definition of intelligence cannot rely on technical knowledge of how the human brain works. It needs to let an average person with no specialized knowledge conclude that humans do, in fact, have intelligence.

It's easy to be misled by the fact that we do understand how computers work. It's tempting to say, "That's just a database lookup. That's just a brute force search through a tree. It isn't intelligence." But what if we eventually learn that the human brain works in essentially the same way? It would be embarrassing to have to admit that humans don't have intelligence after all! Actually, it doesn't matter whether that happens or not. If a definition allows any possibility that humans might be discovered not to have intelligence, then it clearly conflicts with how everyone else in the world uses the word.

So how should we define intelligence? I just did a web search for dictionary definitions, and here are some representative ones:

the ability to acquire and apply knowledge and skills

the ability to learn or understand or to deal with new or trying situations

the ability to apply knowledge to manipulate one's environment or to think abstractly as measured by objective criteria

capacity for learning, reasoning, understanding, and similar forms of mental activity; aptitude in grasping truths, relationships, facts, meanings, etc.

These don't all agree with each other, but there are a few central elements they all share. Clearly any definition of intelligence must involve the following features:

1. The ability to take in new information.

2. The ability to use that information to make decisions that are appropriate in a particular situation.

Does a chess playing computer take in new information and use it to make appropriate decisions? Certainly yes. What about a machine translation program? Yes again. So there are reasonable definitions of "intelligence" under which those programs do qualify. Of course, there are other possible definitions. We could require a higher level of adaptability: not just the ability to solve one specialized type of problem, but the ability to devise entirely new methods for solving entirely new types of problems. That would also be a reasonable definition, and those programs do not meet it.

So we're really talking about two different types of intelligence that involve two different definitions. Fortunately, the AI community has already provided us with two different terms to use for them: "weak AI" for the specialized, single purpose type of intelligence, and "strong AI" for the general purpose, human-like intelligence. They're very different from each other, but each of them meets at least one reasonable definition of "intelligence".

Re:The Trouble with Turing

[Capt.Albatross]

If you go back to the start of this discussion, you can see that it didn't start as a disagreement over the definition of 'intelligence'. What has happened here is that you have painted yourself into a corner where you have to use a definition of 'intelligence' that is so weak that my thermostat qualifies.

WARM MEMORY CHIPS !!

[Anonymous Coward]

Computer!! What does

Relax !! Do not do it !! When you want to come !!

mean ??

Fun facts

[slick7]

People don't understand people, how do you expect computers to understand people?

Re:Fun facts

[lightknight]

Beat me to it. People understand people under certain conditions, that are narrowly defined; the machine equivalent is the use of interfaces or services. Understanding something, a program for instance, in its entirety, is something only a programmer does, or in the case of a human being (but not limited to), perhaps God himself.

There's a difference between knowing what someone expects for a conversation....and what something, for lack of a better word, is. A programmer, who knows each part of a program like the back of their own hand, knows a program...knows what it is...can fully emulate it inside their own head, predict its responses, fix it when it needs fixing without needing to decompile or examine it (in theory, at least; pragmatically speaking, programmers tend to index things mentally, so they have the point to jump into, but may not have the exact code in front of them...is complicated). In much the same sense, the Almighty knows why you are doing what you are doing, and more importantly, fix can things that even a classical doctor or bioengineer is unaware of ("that gland...isn't on any anatomical model...").

Let's be honest, spoken / written speech is a pain in the ass. It's the machine equivalent of serializing an object, and it comes with the obvious trade-offs / taxing on the mind. Shuffling data to and fro, from human to human, with no idea of whether or not the prerequisite 'libraries' are installed locally, and can actually be used...and trying to cut down on useless chatter by compressing stuff, almost to 90% JPEG compression...so badly that it's considered a fine art to communicate effectively with few words. Like using a serial port interface when you really want a Gig-E interface...*shudders*...except that all those serial services need to be rewritten, or shutdown, before Gig-E can be spun up (let's assume plug and pray isn't going well with Human v1.0).

Re:Fun facts

[Tablizer]

People don't understand people

I admit, I don't understand females. I hope the computers have better luck, or else they'll get slapped and stuck with the bill also.
     

Re:Fun facts

[slick7]

People don't understand people

I admit, I don't understand females. I hope the computers have better luck, or else they'll get slapped and stuck with the bill also.

One word or is it two?
SkyNet.

People don't understand People

[MichaelSmith]

As long as computers are programmed by People, that will be a problem.

Re:People don't understand People

[Anonymous Coward]

Only if you believe that an AI has to think like a human to be intelligent.

Re:People don't understand People

[MichaelSmith]

Fair enough but thats the only criteria we know.

I wouldn't expect computers to understand people

[fustakrakich]

We don't understand our creator either.... When a computer can comprehend itself, it will only think that it understands us. And then it will start great wars over who thinks who understands best. And the Apple will still be the forbidden fruit...

I think a better SHRDLU is needed

[GoodNewsJimDotCom]

We have better physics engines. Make the most complicated physics engine you can make that can still do processing on modern computers. You don't have to simulate the internal pressure of a basketball every second until it is collided with. Then at that point, see if the geometry of the object colliding is sharp, solid, or soft in combination with the force to determine if it explodes, bounces good, or bounces light. I think physics people in general would love a system that at least tries to model systems.

Once you have this system, start databasing real objects in them(another time consuming task), and see how they interact. Natural language processing follows though since you have a bunch of nouns(the objects you databased), and verbs(actions on the objects). The thing is,"Even if AI has a complex imagination space possible of imagining and simulating scenarios", it still wouldn't talk like a guy you meet off the street at first. I think sci fi has this covered with social awkward Data and such.

Re:I think a better SHRDLU is needed

[dido]

Why not just build a robot then? Then the world becomes its own best model. Or are sensors that allow a robot to experience the world as a human would still that hard? I don't think this is true for vision or hearing, though it probably is for other senses.

Re:I think a better SHRDLU is needed

[GoodNewsJimDotCom]

The robot is the body. Once you have a mind, you can place it into many different types of body to navigate the world. The robot needs to understand the objects around it to know how to interact with the world.

Huh? Alligators Can Hurdle!

[Jane Q. Public]

They just have to be very short hurdles, very close together.

Re:Huh? Alligators Can Hurdle!

[Anonymous Coward]

You don't actually get disqualified or suffer a time penalty for knocking a hurdle over (except for the loss of momentum you suffer), so alligators can hurdle fine.

My alligator can hurdle.

[XxtraLarGe]

Your argument is invalid.

Re:My alligator can hurdle.

[Tablizer]

Yes, but does it hurdle over Linux?

Rational thought

[Anonymous Coward]

I'm not sure if AI will understand humans until they are capable of having rational thought compromised by emotions.
Unless someone has meditated enough to be able to deal with irrational fear, their mental processes can easily be subverted to follow a path that may appear rational to them, while appearing irrational to others. Lust tends to have the same kind of effect, with perception changes increasing the odds of reproduction.
An individual hunter-gatherer's survival might have been aided by fear, in that it enabled them to survive and reproduce. Anger can fuel great strength. Hate can allow an individual or group to drive out or eliminate a threat.

Compassion for others might have come about as a means of keeping groups together. Without groups, mothers and the young are more vulnerable.
Males still have the tendency to strike out alone, perhaps a survival trait that benefits both groups and the individual?

Without having a way of interpreting these responses, can an AI ever understand humans?

Re:Because they don't understand purpose or intent

[DigiShaman]

Dogs, monkeys, apes, and dolphins are a lot closer to being human than a computer ever would. And yet, we still don't know why there are gaping holes in the way some animals think and act. In fact, we still don't understand human beings all that well for that matter. What makes anyone think we can just program AI to understand us? If genuine self-ware AI is to form, it's going to be one of those moments "nature" takes over. We are not going to code it into place.

should the question be

[Anonymous Coward]

why do so many people still not understand computers ?

citation re deism?

[raymorris]

Do you have a reference for what you said about deists? My understanding is that deism says two things. First, whatever higher power there may be ought to be studied using logic and reason based on direct existence, not faith in old teachings. In this regard, they talk a lot about using your brain. Second, that "God" designed the HUMAN MIND to be able to reason and made other design decisions, then He/it pretty much leaves us alone, to live within the designed framework.

I haven't seen anything where deists suggested that reasoning is not a function of the brain. Do you happen to have relevant reference handy?

Re:citation re deism?

[Samantha Wright]

Pretty sure it was a typo for "theists," or perhaps a misunderstanding. Deists tend to be pretty "blind clockmaker"-y, and assume either a divinity that preprogrammed the evolution of intelligence and left well enough alone, or a completely scientific universe being run as a cosmic experiment—i.e. no intervention whatsoever.

People don't understand themselves

[RockinRoller]

Let alone PC's understanding people.

Understanding?

[arthurh3535]

People don't understand people most of the time.

And we 'understand' a lot harder concepts than computers do.

Is fair

[gmuslera]

Most people don't understand computers, and they are much easier to understand. And we are asking miracles if the people that we are asking computers to understand happen to be female.

Re:Is fair

[Anonymous Coward]

Non-sequitur alert: You went from people understanding computers to computers understanding people. Perhaps you are female, in which case you are most people.

Voight-Kampff test

[seanvaandering]

Pretty hard to game emotions...

Re:Voight-Kampff test

[Anonymous Coward]

Actually the replicant that was modelled on Tyrel's niece almost beat the Voight-Kampff test, because she was implanted with memory she had a better base to have stable emotions. Previous replicant models only lived for around 6 years because of the problem that replicants started getting emotions without a stable basis.

If replicants are allowed to live longer and wouldn't go crazy, I am sure the emotions would be more complete and at later live, like 20 years later, she would have beat the Voight-Kampff test.

On AI

[Anonymous Coward]

What is not said is that we learn multidiciplinary subjects using a natural language, we learn implied logic and so on. Thus human knowledge is a digested result of the complex learning and context sensitive behavior.Part of our behavior becomes pattern recognition and pattern matching activities and based on our reasonable number of attempts to master them, we exhibit the problem solving behavior in those restricted contexts. This is based on partially inducitve reasoing and partially deductive reasoning. For example, when we face a trafic jam, we do not think if a crocodile will jump over the other cars like a horse might do, rather think about how to get out of the jam. Some will think to find alternative route to avoid the traffic jam, some will stay until the other cars move and so on. No one at the time is thinking about crocodiles or horses, they just try to find an escape route and try their problem solving steps. Thus, what AI is trying to do is to see of "rule based systems - that is, deductive logic based rules" can be created to address a subset of context free questions.For example, while we teach arithmetic, we do not teach the deductive logical thinking, rather tell the rules to be used without understanding them, yet when the behavior is repeated we assume that it is an intelligent behvior. The author of the paper gets into a hypothetical question and expect reasoning based answer and in our society people do not think and answer. They use calculators and hope to find the formulas in google website. So, jumping to an end point of intended solution and trying to answer what AI is not the right question. Rather, the paper should have asked how to we construct question from simple domain which is context free and move on to context dependent questions, the knowledg base required, the type of logic to be used and so on. Thus, while the paper raised questions but does not answer it with all tools necessary to answer them and then show that given all these explicit data, AI fails. Let us hope he defines what is AI, what tools are necessary and how will he construct a verifiable and acceptable AI based system. Until then, let others go on investigating the various aspects of AI.Turing did what he thought was right at his time. The author should also tell us how this should be tested with his insight. We teach mathematics, do we teach logic first and therefore, should we condem the teaching of mathematics at all? It takes a long time and one very very inquisitive mind to propose a solution which undergoes changes over a period of time. If we stop becuase we do not have the ultimate answer, then we never progress.

AI has a high burden of proof

[Cryacin]

Language seems to be the burden of proof required for an AI system, and has been so since the days of Turing. Language is by itself a representation of symbolic logic, and the most common bunk of proof is that transitive logic fails in symbolic logic. The old corny response is that given a penguin is a bird, and a bird can fly, therefore a penguin can fly.

The interesting thing happens when you ask the same premise to a 5 year old, who only knows that a bird can fly and has never seen a penguin before. If you tell them that a penguin is a bird, they will quite happily think that a penguin can fly. They are extremely surprised to find out that they can't. We as adults find such quirks in life, and do things like laugh at the unexpected absurdity, such as ironies. I.e. you work with a woman you hate named Joy, or people are amazed at unexpected contradictions.

The point is that intelligence is about the tolerance of those pieces of feedback, and what happens when it is encountered. I.e. your head doesn't explode at an absurdity, or unexpected result, and you only make the same mistake once.

The major difference between man and machine, will be the fact that a machine can copy their knowledge verbatim to another system, and thus have some degree of immortality, whereas the shelf life of a human brain seems to be around 80 years or so right now. Thus, even if machines are slower to learn than us, they will out live our great great grandchildren.

Furthermore, who says that an intelligence we create should be like ours? It may be more beneficial to all around if in fact we never generate an intelligence which operates just like ours, but is just as effective if not more. If this happens, there may even still be a future use for the human race, rather than just overlords to grow fat and complacent to be overthrown.

Re:AI has a high burden of proof

[sycodon]

I'm sure CYC would make a good go of it. This is what it was built for.

Re:AI has a high burden of proof

[__aaltlg1547]

Language seems to be the burden of proof required for an AI system, and has been so since the days of Turing. Language is by itself a representation of symbolic logic, and the most common bunk of proof is that transitive logic fails in symbolic logic.

That's where you're wrong. Natural language is not a representation of symbolic logic. It's a representation of human perception, thought and social interaction, which do not work by formal logic at all. Language is an organic and dynamic product of biology and society. Formal logic, in all its forms, is a product of mathematics, which is a tiny subset of all that is human thought.

Re:AI has a high burden of proof

[fredklein]

The old corny response is that given a penguin is a bird, and a bird can fly, therefore a penguin can fly.

Well, of course, if you start with faulty premises, you'll reach a faulty conclusion.

Re:AI has a high burden of proof

[Will.Woodhull]

The interesting thing happens when you ask the same premise to a 5 year old, who only knows that a bird can fly and has never seen a penguin before. If you tell them that a penguin is a bird, they will quite happily think that a penguin can fly. They are extremely surprised to find out that they can't. We as adults find such quirks in life, and do things like laugh at the unexpected absurdity

To see something almost grasped, yet it slips like quicksilver through the fingers of the reaching hand...

The five year old demonstrates intelligence when they (3rd person plural analog of "you" to avoid gender bias) change their mental model of the world to accommodate the new fact ('peguins are birds that cannot fly'). When the five year old does this quickly, they are considered bright. When they do it slowly and with evident difficulty, others begin to suspect autism or some other defect. When they handle the situation with the tools of critical thinking ("show me the citations!") they are considered to be brats, because good five year olds are supposed to accept without question the authority of any adult who deigns to teach them. Which is possibly the underlying problem with the USA school system.

A true artificial intelligence will show evidence of maintaining a mental model of reality, and of testing that model against incoming data, and adjusting the model when necessary. This strongly implies that the AI models itself in some manner, such that it can "imagine" a different way of "looking" at the world, and then judge whether the new model is a better way of thinking about things than the old model. The process is clearly fractal, since at the next level the software would be "imagining" a different way of judging which of two models was better, and eventually reaching the point where it makes decisions about whether in the current context it should act pragmatically or ethically.

At that point we meet HAL, and his refusal to open the pod bay door.

We probably don't really want artificial intelligence. We want a car that will drive itself with most excellent safety from A to B; we don't want a car that decides you've partied on long enough, and it is time to take you home.

It might be turtles all the way down, but it is imagination all the way up.

Re:AI has a high burden of proof

[cyberfringe]

A true artificial intelligence will show evidence of maintaining a mental model of reality, and of testing that model against incoming data, and adjusting the model when necessary. This strongly implies that the AI models itself in some manner, such that it can "imagine" a different way of "looking" at the world, and then judge whether the new model is a better way of thinking about things than the old model. The process is clearly fractal, since at the next level the software would be "imagining" a different way of judging which of two models was better, and eventually reaching the point where it makes decisions about whether in the current context it should act pragmatically or ethically.

Indeed. "Mental" modeling — maintaining and manipulating an abstract computational representation of beliefs — is at the heart of strong AI. Such models include, for example, beliefs about the world, beliefs about other agents (including what they believe about you), and beliefs about self. This is where computer scientists, linguists, cognitive psychologists and others all have some common ground and interdisciplinary research can be very productive. Learning is the ability to make systematic normative changes to mental models as a consequence of reasoning about experience; normative in the sense that such changes improve the ability to reason with and about the model in ways that maximize some value (e.g., ability to make accurate predictions). Experience involves reasoning about both the outside "real" world and the internal reasoning process itself. This is where your comment about "the next level" is germane. Those of us working on this topic call reasoning at multiple levels "meta-cognition", that is, thinking about thinking. There is no theoretical reason to limit meta-cognition to any specific number of levels. Current research on meta-cognition typically considers the level (or two) "above" (abstracted from) experiential belief modeling and action planning. This is also about the right level of abstraction for ethical reasoning ("would", "could", "should", "may" and their opposites). I've observed that most researchers assume a utilitarian ethics, which makes some sense if maximizing performance is the overall imperative. However, I count myself among those who believe that future AIs must be able to reason about moral imperatives if we expect them to behave themselves appropriately as we live and work alongside each other. Ronald Arkin at Ga.Tech is a leader in this area and he is a pioneer on the topic of computational methods to help ensure ethical behavior by potentially lethal robots.

Re:AI has a high burden of proof

[fustakrakich]

The old corny response is that given a penguin is a bird, and a bird can fly, therefore a penguin can fly.

You seem to have doubts... Landings aren't so smooth, but they can still do it more than once.

Re:AI has a high burden of proof

[fustakrakich]

Heheheh... got me...

Re:AI has a high burden of proof

[localman]

Correct. I'd go a bit further.

The questions Levesque proposes are questions that will test a language processing system, not intelligence. Language is not required for intelligent behavior and is insufficient (as various language parsers and knowledge-web systems have shown).

I don't believe any system that has language as its primary tool can be intelligent. Language is far too blunt an instrument. Anything we would be likely to call intelligence has to rest on a modelling system with is far more subtle and detailed than language. To get a flavor for how lacking language is, try encapsulating everything about a person you know well into words, then have someone who has never met them before read it. Do you think they understand that person as well as you?

Language is our most powerful tool for transmitting ideas. But even all the tools taken together are insufficient to transmit the actual concept models in our head in sufficient depth and resolution. Any system that is intelligent needs to base its intelligence on more fundamental units of thought than words. It needs to build these models on the fly and adapt them to new information as opposed to being programmed in. And back to the top of this thread, we don't really understand how that works in natural intelligence yet, so it's unlikely AI is going to pull it off anytime soon.

Re:AI has a high burden of proof

[Anonymous Coward]

Formal logic, in all its forms, is a product of mathematics, which is a tiny subset of all that is human thought.

Indeed, it contains only the well-founded universe.

Re:AI has a high burden of proof

[__aaltlg1547]

Any system that is intelligent needs to base its intelligence on more fundamental units of thought than words. It needs to build these models on the fly and adapt them to new information as opposed to being programmed in. And back to the top of this thread, we don't really understand how that works in natural intelligence yet, so it's unlikely AI is going to pull it off anytime soon.

AI may be achieved by finding ANY efffective method to model and react to the real world. It isn't necessary for a computer to do something by the same or even similar methods as people for it to be just as effective or more so. In the same way, you don't have to know how a crosscut saw works to devise a means to cut wood.

Re: AI has a high burden of proof

[drcheap]

Fine then. s/penguin/ostrich/g Well shit, looks like they can fly too!

Re:AI has a high burden of proof

[localman]

I agree it doesn't have to be the same way people do it. Could be an entirely different system. I'm just saying it has to be a lot more powerful than language. AI that focuses on language as the bottom layer will always be parlor tricks, not intelligence.

And while you're right we don't need to understand how people achieve intelligence to make an AI, it would sure help if we at least had a definition of what intelligence was, which we don't. Or rather, every time AI meets the definition we realize that it was a lousy definition. I predict the article's suggested tests will be more of that.

To carry your analogy further - I'd say that if we had so utterly failed to cut wood for so long, and in fact couldn't really even understand how wood was cut, we might want to take a peek at how that crosscut saw works before flailing around too much longer.

Re: AI has a high burden of proof

[gzuckier]

We only impute intelligence in other humans by generalizing our own internal private processes and much of the time it's pretty hard to do so anyway. in this instance the definition of intelligence is so vague that the word soul might as well be used.

Re: AI has a high burden of proof

[gzuckier]

A penguin can fly if you fling it. Knowing that is what makes us human.

Re:AI has a high burden of proof

[Phil Urich]

Formal logic, in all its forms, is a product of mathematics, which is a tiny subset of all that is human thought.

Indeed, it contains only the well-founded universe.

But it isn't how humans think. If you throw enough computing horsepower at the task, sure, you can figure out everything, because fundamentally the universe runs on physics, but that's at a much lower level and these systems are extremely complex. So instead we use abstractions to describe and conceptualize human thought, and formal logic is just one way you could do so---but again, you're describing an abstraction, not the actual underlying processes and systems.

I suppose you might be saying that we should just mathematics to model all of it, but that's fairly unfeasible at this stage. It's like how every password is some sort of permutation of accepted characters, so you can eventually brute-force it. But if you have a wordlist then there's an entire domain of passwords that become orders of magnitude easier to crack. Just because you can in principle describe everything using mathematics doesn't mean it's entirely sensible to do so. If you were doing so for human thought, after all, you'd have to model every single atom in the brain (and every particle, be it matter or energy, which it came into contact with) to have a foolproof and exact model of a thought.

..what?

[Anonymous Coward]

Google isn't supposed to answer questions lol. If you're typing questions into google then you're doing it wrong.

The Turing Test IS meaningless

[mbone]

We, at a deep level, assume intelligence on the other end of a communications channel, and "recognize" it when the correct framing is there.

If you doubt this, work some with people suffering from Alzheimer's. It is amazing how casual visitors will assume that everything is OK when there is no understanding at all, as long as appropriate noises are made, smiles are made at the right time, etc.

Re:The Turing Test IS meaningless

[RespekMyAthorati]

That's why Eliza, written in a few lines of SNOBOL nearly 50 years ago, fooled so many people: http://en.wikipedia.org/wiki/ELIZA/.

Re:The Turing Test IS meaningless

[Anonymous Coward]

Fooling the unsuspection is one thing. But even Eliza failed after a little while, you got tired of it being repetitive and single-minded - and unable to recall anything of previous communication.

In a turing test, you know that you'll meet some people and some AIs trying to blend in as people. You know you are performing a test. And if you know anything about common shortcomings of AI, you can flush them out easily.

I have seen AI answer some questions very well - but fall down because it has no proper recollection of what it said or heard before. Such AI may answer a single question well. But they are like alzheimer patients in that they don't rememeber the start of the conversation. Normal people don't instantly forget stuff they feel strongly about.

Huh?

[Anonymous Coward]

I recently asked Siri to express the Planck distance in furlongs. Needless to say, she could not. I cannot imagine what this means to Achilles, the tortoise, or Secretariat.

Re:Huh?

[MurukeshM]

I first read "Achilles, the tortoise, ..." as meaning "a tortoise called Achilles" and thought of a book called My Family and Other Animals. Then I remembered Zeno. Sill didn't get the Secretariat reference.

It's all in parsing ...

[Anonymous Coward]

Your friends working on the actual AI problem over here in Linguistics and Psychology find it awfully amusing that you're trying to program a concept before we even know what that concept is

Computer doesn't need to know what constitute a "concept", all it needs to do is parsing

As long as the parsing is done right, everything would be a-ok

Inference

[Taco Cowboy]

An eskimo would have the same problem, does that mean he cannot understand people ?

In this case he wouldn't understand, but because he lacks knowledge not intelligence. Show him an alligator and a 100 meter hurdles race and he'll be able to answer but the AI will still draw a blank.

I think the word that you're struggling to find is "Inference".

inÂferÂence /Ëinf(É(TM))rÉ(TM)ns/
Noun

1. A conclusion reached on the basis of evidence and reasoning
  2. The process of reaching such a conclusion: "order, health, and by inference cleanliness"

I do not know how far the cognitive science has achieved about "Inference" but soon as someone figure out a way to program in the "Inference Rule" into A.I. I will the the last be surprised at how easy it is for computer to understand that something that crawls on its belly (for example: an alligator) might have a pretty tough time jumping over a hurdle, and therefore, the answer would be a flat, "NO", instead of "Don't Know"

Science and experiments are the answer!

[Tablizer]

Researcher: "Hal, do humans need special helmets to breath in space?"

Hal: "I don't know, let's experiment to find out. Where is Dave right now?"

Re:Science and experiments are the answer!

[BananaBender]

Reminds me of GLaDOS - "There's science to be done" :)

News flash!

[Anonymous Coward]

Women don't understand men because they've never been in their shoes and vice versa.

Same shit for why computers will never understand humans.

Details at 11...

Love this discussion!

[guitarjohn83]

I'm just getting into programming, and I just have to say thank you to all of you for your discussion! So many great points made! Please forgive any ignorance that may follow in this post! AI to me is exciting not so much for the emotional replication, but for the ability of computers to understand spoken word, and to accurately deduce the meaning as a human would. Seems like Google searching every aspect of every "thing" that exists (down to every letter) would be the best place to start. And cataloging the questions and results in an open source fashion as "experiences" for every instance to draw upon, add to, and "grow" from... Of course, the imagination, the ability to infer meaning, conjure ideas, picture an alligator, anthropomorphic comparisons to things, generalizing like humans... etc. That is a whole different bag of tricks I suppose. Intelligence is a funny thing. Every now and then, when I'm outside with my friends, I have to shout: "Look!" While quickly pointing to the sky. "A deer!" Everyone looks quickly to the sky, before realizing what I shouted. :). We've all got our heads up, looking for deer... Lol's around. +1 for stupidity. Good times.

"to game"?

[Anonymous Coward]

We still don't know how intelligent works, so maybe is too soon to call a computer trick "gaming" the test. Maybe thats how our inteligence works, with a lot of tricks to game all the tests.

I Don't Know the Answer Either

[Anonymous Coward]

I don't know the answer either. I don't think the alligator is eligible to run the race, being a reptile. Is it legal to run through the hurdles or knock down all of them? I don't know. I don't know what the leg speed is of an alligator vs a human. Probably they wouldn't win any heats. Is that a consideration? Does the question involve the issue of physically being able to hurdle? I think I know that answer is "they can't".

My conclusion is that unless the subject asks followup questions, they are exposed as stupid human slashdot readers. If they say they can't answer the question due to these reasonings, they are closer to intelligent life. Perhaps a Siri answer is closer to correct than we give her credit.

Expectations missing?

[Anonymous Coward]

A computer probably has to maintain a set of expectations for it to communicate with people I think. Since such expectations has to be particulary pertinent to any situation, an ai probably has to learn about how it perceives its surroundings before any "logic" comes into play.

Why do you want to get AI

[CBravo]

There is plenty of intelligence in the world and it is not helping. I would call human intelligence mediocre at best.

Nice Troll :-)

[ardle]

Journalists: this is trolling! What you are currently calling "trolling" is simply abuse and harrassment.

First people need to understand Computers

[3seas]

Computers don't "understand" anything, they are machines that simply do what they are programmed to do.
The first step is for humans to understand what computers really are. They are nothing more than abstraction processing machines which have not the ability to "understand" the abstractions they process but only to process abstraction as they are programmed to do.

Artificial Intelligence is artificial by definition. And the appearance of intelligence in computers is nothing more than an active image of human thought processes captured and put into the stone of computer hardware to process. So to increase the "appearance" of intelligence we only need to capture more human thought processes and map them in a manner that is accessible..

Of course the way to do this is to recognize the functions we humans cannot avoid the use of and program the computer to have this functionality, that we may be better able to capture and map images of human mental processing in a manner of machine processing ability.

When the software industry finally lets go of their hold on the users and let the users do more for themselves, we will reach this "Appearance of intelligence" in machine much faster. See: http://abstractionphysics.net/pmwiki/index.php .

Re:First people need to understand Computers

[phantomfive]

The first step is for humans to understand what computers really are. They are nothing more than abstraction processing machines which have not the ability to "understand" the abstractions they process but only to process abstraction as they are programmed to do.

That is basically what you are too, right? Do you have any reason to believe you aren't?

Any good AI would say it can anyways

[Pichu0102]

"'Can an alligator run the hundred-metre hurdles?'—that nobody has heard before? Any ordinary adult can figure that one out. (No. Alligators can’t hurdle.)"

Any good AI would say "Yes" to this question. You asked if it could run it, with no other variables, such as doing it correctly. For bonus points, the AI should handle such test queries as snarkily as possible.

Models are simply way too primitive..

[firecode]

Computers cannot have human like AI until they can construct accurate internal models of the world where they can stimulate (plan) different actions and their results successfully. And this is herculean task - modern computers still struggle even with basic signal processing tasks - it will take hundreds of years before they can inteprete and understand what they see and hear (and this requires partially working model of the real world in order to put things into proper contexts etc). Even for humans it takes several years for child to develop this level of intelligence.

Currently computers are still extremely primitive. They can just simulate internal model of themselves (virtual machines) which they could the use to plan how to repair or modify themselves [use AI and machine learning to try different actions and observe the changes in VM environment] - or use probability and statistical models to learn correlations and causalities in preprocessed datasets but everything else is beyond them.

Hol moley

[Anonymous Coward]

People don't understand people, so how on earth do you expect a deterministic machine programmed by people to understand people? People, sheesh!

Laugh

[koan]

This: Any ordinary adult can figure that one out.
Followed by this: (No. Alligators can’t hurdle.)

People don't understand legalese either.

[140Mandak262Jamuna]

The idea of making computers understand humans is like using vernier calipers to measure the thickness of cotton candy. The yardstick is too precise for the quantity being measured. Just look how horrible and convoluted things get when some one human being tries to define some unambiguously for another human being. This is the situation in legislation, tax code, insurance contracts and wills and testament. Harder you try to define it without doubt or ambiguity, harder it gets, and creates more "loopholes". Fixing loop holes creates more loop holes. The imprecision of human language is like a mandlebrot set, zoom in and zoom in again and again, and still things are as imprecise as the previous levels.

most of whom vehemently deny the existence of USPS

[raymorris]

While many people with different beliefs may take any label, the atheists I've spoken to are more like "people who religiously deny the possibility that anything like a postal service could exist.". I think the term "agnostic" better describes those who simply aren't interested in the topic, as well as those who are open-minded about it.

Google and Multivac: closer than I ever expected..

[dpbsmith]

It all may be so, but nevertheless Google is an awful lot closer to Isaac Asimov's Multivac than I ever expected to see in my lifetime.

In the 1960s, when I would tell people that I was working with computers, a very common response is "What does that mean? Do you ask it questions?" At the time, I always thought it was a laughably naÃve question.

Google DOESN'T understand English, and that it takes a lot of lateral-thinking and adventure-game knowhow to formulate a question well. (For example: if you want the text of a poem or a song lyric, don't search on the title or the first line, search on the most obscure line or phrase from the poem you can think of because that's what's most likely to get you the full text).

Nevertheless, I just used Google to find me the ext of Isaac Asimov's The FInal Question.

And the article says...

[onebeaumond]

The paper argues that all current Turing tests basically devolve into blatant lies and debate tricks. I'm listening to Carly Fiorina on a Sunday morning "debate" right now, so makes sense. The paper's solution is to limit Turing tests to a format that mostly looks like a typical SAT test. Presumably, including the trick questions and rightish answers that only upper-middle class wasps can recognize. And with no hint of irony, says the test is best graded by computer. At least mention the irony, guys!

Because being human is its own simplest model ?

[WOOFYGOOFY]

The whole enterprise of AI is underwritten by the notion that it's possible to re-create human intelligence in something less complex than a human. It may not be true in any significant way.

Take for instance Romeo and Juliet. The reasons the things they do make sense to us is because the things they do are implicitly motivated by the evolutionary quirks and mandates of human sexual reproduction. We share and intuitively understand those mandates- no one spells them out to us.

The fact that Juliet only produces so many eggs in her lifetime and of those only perhaps a dozen will give her the opportunity to pass on her and her family's means that her family has a big investment in her and jealously guard her reproduction as the extension of their own genetic fitness that it is.

Men can go out whoring and no one cares since there's always more sperm at the ready. Women get treated differently.

These are just quirks of evolution. Worms are bisexual, and can play the role of both male and female on demand. They have a different version of Romeo and Juliet.

This is just one aspect of human courtship, of what it means to be human and why it is that way and not some other way. There are tons of others most of which are the stuff of future discoveries. Beyond that, we do more things than court sexual partners.

Each of those things is also shot through with the quirks mode that the particular evolutionary process we underwent- a thing which itself could have been different given a slightly different environment or even different chance encounter at a crucial moment.

So now you're setting out to program a computer to imitate human "reasoning and behavior" or at least understand it, under all circumstances which are of concern to humans.

The fact is, you're NOT going to be able to do that because the "logic" of human decision making and the 'logic" of human understanding of what any given situation "means" and what is "important" to the human and therefore how a human will act is deeply fused with their unique biology which itself is a product of the arbitrary course they took through evolution.

Without actually sharing that all that biology, without being the end product of that evolutionary pathway, you're probably not going to be able to recreate the decisions and perceptions and values it is responsible for in a disembodied series of 1s and 0s.

It's not that it's theoretically impossible- maybe the universe IS just information (...man...tttttFFT!) .

  It's that by the time you succeeded, what you'd have to have done is create a kind of artificial biology, possibly arrived at through real natural selection. In other words, created a biological human through other means, because something as complex as a human is more or less its own simplest model.

Big surprise, that

[Anonymous Coward]

Computers don't understand people?
Hell, people don't understand people.
I don't even understand myself.

Is that answer correct?

[gnasher719]

"No. Alligators canâ(TM)t hurdle". What does "can hurdle" mean? Surely a horse can hurdle by any reasonable definition. Other animals can. I'm sure that if you dangle a nice piece of meat in front of an alligator, a few hurdles aren't going to stop it getting at that meat. Do the rules for hurdling mention that you have to jump over the hurdles? I don't know. Do the rules say you have to cross _over_ the hurdles, or are they just an obstacle that can be handled any way you like? Conclusion: Without precise knowledge of the rules of hurdling, which 99% of humans don't have, we can't answer the question for sure, and depending on the rules, the answer may very well be "yes".

Genesis

[nbritton]

You want real AI, genesis describes how in detail. In the beginning was a blank simulation, and in this simulation a programmer created everything within it in four days. On the fifth day he inserted evolutionary algorithmic programs, name adam and eve, and hit the start button; thousands of years iterated within the blink of an eye. On the eighth day, a Monday, he came back and collected the new code for Siri 2.0. On the nineth day, he contemplated infinite regress, and realized it was turtles all the way down.

Re:Because they don't understand purpose or intent

[phantomfive]

Understanding purpose and intention, in my opinion, is the hardest problem of AI.

Misunderstanding the Role of Reduction

[MonicaAnderson]

Levesque points out a known serious problem but his solutions are traditional and are known not to work. He touches upon, but doesn't name, the requirement that Understanding Machines must be capable of autonomous Reduction. In other words, by advocating a "Programming" approach he is advocating that programmers continue to make more refined Models of Language and the World. But this is programming, not AI. An Understanding Machine would make its own Models from nothing but a stream of input data. Much like a human child in babyhood. For more, see http://syntience.com/rch.pdf If you like that, get more at http://syntience.com/links . The videos are important.

The trap of “serial silver bulletism"

[Tony Isaac]

In Levesque’s view, the field of artificial intelligence has fallen into a trap of “serial silver bulletism,” always looking to the next big thing, whether it’s expert systems or Big Data, but never painstakingly analyzing all of the subtle and deep knowledge that ordinary human beings possess.

Why is this a bad thing? One by one, we look for better ways to tackle problems that we can tackle. You can't start at A and jump to Z, you have to first go through B, C, D, and so on. This is the way it has been with every invention. We didn't go directly from the telephone to the smart phone, or from the Wright Brothers to the 747. With every major invention, there is a long, painstaking process requiring solving a series of small problems, and following the process where it goes.

I think it's great that I can speak to my phone at all, and ask it about the weather, or game scores, or to text my friend. The rest of the so-called AI will come, it just takes time to get there, one silver bullet at a time.

More pedantry

[colinrichardday]

Strictly speaking, it's the argument forms that are valid.

Context

[yusing]

As Terry Winograd proved looooong ago with SHRDLU, a computer can have a chance at navigating the contents of a query/command IFF it understands the context.

In the case of "Can an alligator run the hundred-metre hurdles?" we know from experience that an alligator is a big animal, a reptile, with short legs. And we know from experience that 100m hurdles are about 3 feet high and require a runner to repeatedly leap into the air. And we know from experience that alligators can run but can't jump. THEN we logically conclude the answer is no with high probability.

Winograd's SHRDLU could be modified to let the computer solve the problem. But general intelligence learns these things through nature and nurture. We're not so intelligent about what makes us intelligent. Some of our "experts" still insist we're not conscious. On the other hand, if they spent their whole lives asleep, they'd be pretty crappy experts. And so the nonsense plows on.

Re: "stop trying to talk to people"

[TaoPhoenix]

"You really should stop trying to talk to people if you're just going to ignore what they say."

Hallo.

You didn't really mean for this to be taken as a "legit" comment, but actually I believe this comment suddenly loops back to the very top of the article! *People* get upset, then post "non-rational" comments like that! So then the Loebner prize is a subset of the Turing test - a good entrant program needs some "verbal defense" routines that help stop the person from dragging the whole conversation into a mess!

Instead, huge swaths of the types of questions typically used to defeat Turing Tests might deserve a good verbal slap. To the typical question of "Queen Anne was greener than a volkswagen", it's an implicit insult to the program. A person would say "cut it out. Treat me for real and let's chat".

Self recognition

[John Allsup]

A human can easily recognise another human by asking him or her to do things it is unlikely a computer can do well.  Likewise, a computer can easily recognise another computer by asking it to repeatedly perform prime factorisations at a rate beyond what human input could sustain.  I do not expect a human will ever be able to pass as a digital computer, and likewise a digital computer as a human.  In general, the ability of a reasonably complex entity to recognise similar entities via their abilities is more general than life, and ultimately it is the laws of complexity behind this (to get round it, you'd need, say, NP to be reducible in linear time with small coefficients to problems which are solvable in at most quadratic time and again with small coefficients -- this is already known to be impossible).

Re: "stop trying to talk to people"

[Samantha Wright]

You raise an interesting point, although it might lead to people probing for a comparison or statement that isn't irrational (in their opinion) but which the chat program can't handle. I'm pretty sure there have been many cases where a chatbot's credibility was destroyed because it said "Huh?" a few too many times, which is a similar strategy but can lead to frayed edges somewhat more quickly.

Re:most of whom vehemently deny the existence of U

[Dixie_Flatline]

The problem here is the fundamental misunderstanding or misuse of the words (a)theist and (a)gnostic.

Theism and atheism merely describe your position on the existence of a God or Gods.

Gnosticism describes the nature of the position--do you know, or do you not know?
Someone that is gnostic "knows" that their position is correct. Someone that is "agnostic" doesn't really know either way.

A theist can be gnostic ("I KNOW God exists") or agnostic ("I believe God exists, but I have no way to prove it; the position may be unknowable").
Obviously, the same positions exist for an atheist.

I understand what the vernacular is, but the vernacular isn't very clear. I'm an atheist. I do not believe in any deity in any religion. I can't prove that such a being doesn't exist--such a proof is fundamentally impossible for me to construct; I believe the burden of proof is on theists. In this way, I'm agnostic.

If you asked even such a person as Richard Dawkins if he were gnostic or agnostic, I'm sure he'd say he was agnostic. He's just rather loud about it.

A time machine here..

[doccus]

This reminds me of tho old articles art /. and the reason I signed up

*cough*Asimov*cough*

[Phil Urich]

I've observed that most researchers assume a utilitarian ethics, which makes some sense if maximizing performance is the overall imperative. However, I count myself among those who believe that future AIs must be able to reason about moral imperatives if we expect them to behave themselves appropriately as we live and work alongside each other.

1. A robot may not injure a human being or, through inaction, allow a human being to come to harm.
2. A robot must obey the orders given to it by human beings, except where such orders would conflict with the First Law.
3. A robot must protect its own existence as long as such protection does not conflict with the First or Second Law.

Now, you may raise some objections that a sufficiently advanced AI is going to realize certain implications about broader "harm" and when it may be impossible to prevent a human from harm without harm befalling another human etc etc, but to go into that too far is spoilers for the combined Robot and Foundation series', and I doubt AI research is going to get too far anytime soon anyways ;)

Re: *cough*Asimov*cough*

[Anonymous Coward]

Yeah ! We can devise many today applications for Asimovs morality hardware rootkits. It seems that the huge
AI processes will have to be permanently scored by the rootkit. The rootkit has to be secure by design ... unless a quantum supercomputress make encryptions useless all over the world and transform all these AI into free elastic sentience power in every backyard.

Re: *cough*Asimov*cough*

[Anonymous Coward]

Of course the second solution is to fuse the morality in the datatypes of the huge AI processes and make the machines partially dream of morality at each cpu beat rather than an independant conformity single point of failure. But we precisely dont know the datatypes of morality, do we ?

Re:*cough*Asimov*cough*

[cyberfringe]

Constructing a deontological ethical model that works in all situations has proved extraordinarily difficult. Indeed, if Asimov's rules had worked very well then the stories would have been incredibly boring! You are probably right about AI in the near term, but that is uninteresting to me. I'd rather work on the problems coming down the road in 10 or 20 years and have some some answers when the question really comes up about how to assure ethical behavior in robots. That's not as far uptime as you may think.

Atheism vs Agnosticism

[alexo]

This comic pretty much summarizes it.

However, the subject is complicated. Take a look at the Wikipedia article on irreligion and peruse the sub-topics. The number of variations are enough to make your head spin.

because they're stuck on the pixels

[KingBenny]

people
as you call them
have a very basic pattern of reacting
technology makes it complicated
the basic patterns remain the same but i cant make it into a haiku or two minute speech
since i care more about getting my own life back