Slashdot Mirror
Understanding an AI's Timescale
An anonymous reader writes "It's a common trope in sci-fi that when AIs become complex enough to have some form of consciousness, humans will be able to communicate with them through speech. But the rate at which we transmit and analyze data is infinitesimal compared to how fast a computer can do it. Would they even want to bother? Jeff Atwood takes a look at how a computer's timescale breaks down, and relates it to human timeframes. It's interesting to note the huge variance in latency. If we consider one CPU cycle to take 1 second, then a sending a ping across the U.S. would take the equivalent of 4 years. A simple conversation could take the equivalent of thousands of years. Would any consciousness be able to deal with such a relative delay?"
I hope they are nice to us.
---- The above post was generated by the Turing Institute. Maybe.
AI would form its own Fermi Paradox : If there is intelligent life , then why aren't they answering ?
Avoid your fears , or wonder at the past
One CPU cycle as one second might be a good metaphor for computer memory but not AI. It's closer to the equivalent of a neuron firing in the human brain, then it is to 1 second of human time. Human speech takes more than one neuron to fire, and it would take way more than one CPU cycle to process. An AI algorithm which is processing data, and analyzing it would literally take millions or billions of cycles most likely to do the most basic things. While no doubt speech recognition has gotten much faster, it is still and probably will always be a massive undertaking for a CPU to do, as opposed to say adding two 32-bit integers.
Read Dragon's Egg by Robert L. Forward. (and the sequel, Starquake) Part of the story involves humans interacting with an alien species that is a lot faster. The alien's lifespan is about 15 minutes...
Of course an AI consciousness would be able to deal with such a relative delay... They wouldn't be very "intelligent" if they could not. Duh!
No task can be accomplished in a single CPU cycle.
A human can actually do something in a second, like move or talk.
Seriously, is speculation about how bored AI might get, y'know, if it actually existed, really worthy of a /. discussion? I mean, it's a bit like solving about how guardian angels stay warm when flying to help people on the North Pole.
Artificial intelligence at the level of human consciousness doesn't actually exist. Any technology that could create/sustain a true subjective, intelligent experience wold have to be so complex that I suspect managing the perception of time as it relates to human perception will be the least of their concerns.
One CPU cycle = 1 second perceived is totally ridiculous as a premise for making such a comparison. What is the our perception, time wise, of a single neuron firing?
To a computer time is meaningless; you can 'suspend' a program and resume it. Pop data onto a stack and pull it back later. It doesn't 'age', there's no lifespan; in fact even if that hardware from 30 years ago completely dies, I can load it into an emulator. I turned on a computer from 30 years ago. it runs just fine, it can even connect to the internet.
Furthermore, a consciousness in a computer would have to deal on these timescales in order to survive and be meaningful to us; Such an intelligence that didn't learn to deal on these timescales would not survive (thing maintenance intervals on machines, shutdowns at night/weekend, etc). So sure it may 'exist', and even last for billions of its cycles, but if it cannot persist past these thresholds, its irrelevant; much like an animal in a tidal pool that dies before the tide comes back; the ones that made it past that were our ancestors.
A 'cycle' doesn't constitute a thought. I would be willing to bet that a human brain can actually process speech faster than a computer can. (not sure how you'd prove that.)
Computers aren't sentient NOW because they aren't fast enough yet. At least, that's a staple of science fiction. It's only when the computer gets 'big' enough...gets 'fast' enough that they can start to be sentient. So saying when a computer becomes sentient it will suddenly "think/talk" magnitudes faster than us is a non-sequitur.
Now, what they will have is photographic memories. They'll have a huge advantage in the 'random access memory recall' area. I assume it's possible they'll be better at 'hand-eye' coordination. (Not that she had any hands in 'her'.)
--Welcome to the Realm of the Hawke--
Our brains are electrochemical, and don't run that fast in series, but they're massively asymmetrical and parallel, and running in async. We don't have clock strobe lines or addressing, and it might take millions of internal connection points to provide a fast enough interface. Even that might be six orders of magnitude slower than our AI buddies.
The retina and optic nerve might be the closest we might get, so are you willing to give up an eye for this? Go full-on Borg?
John
Just as our minds are not points of awareness but collections of such, any AI will have processes that evaluate information at differing timescales. Moreover, the consciousness of an AI will be at whatever timescale the AI most commonly needs to interact at to thrive. All other processes will be subordinate, whether faster or slower.
A AI, a True AI would set aside a fraction of its self to "talk" to humans.
"If any question why we died, Tell them because our fathers lied."
Well then it just has to spin off a copy of the AI onto a long running thread which just sleeps between the "1000's of years" equivalence of communicating with a human. If it is sound asleep time is not an issue :p
OP's entire premise is pretty thin.
Human beings perceive light, for example. (They can also perceive electricity, to a degree, but that is not as relevant to the point.)
But while a human being might perceive that a flashlight at night has shined his/her way, it takes the same amount of time, roughly, a a fiber optic signal from the same distance. So what?
Generally, it is the speed of perceiving and interpreting the signal that takes time, not the speed of its propagation. We communicate at lightspeed, too. Or close to it. Anybody who has had a video chat has done that. Did that make you superintelligent?
We have never built an "AI". And in fact we have NO reason to believe -- no evidence whatsoever -- that its speed of perception and interpretation would be any faster than our own. There is a very good chance that it would be much slower... at least in the beginning.
I would like to remind people that the idea of "intelligent" machines has been around for almost 100 years now. AND we still don't have any solid evidence of being close to achieving such a thing. Sure, computers can do a lot, and what they DO accomplish, they tend to do very fast. But what they accomplish is not "AI". Even Watson is not "intelligence", it is only the illusion of it.
Even if computers manage to develop a conciousness, and that conciousness have anything in common with human ones, in particular regarding motivations (2 wishful thinking hypothesis with probably little ground behind), what will be its perception of time? Is not just a cpu cycle, our individual synapses goes far faster than our perception of time, and if well computer cycles are faster, their emulation layer toward building a neural network as complex as human one may be far less efficient.
This article at Science Daily is helpful in understanding the issue: http://www.sciencedaily.com/re...
Comparing CPUs and brains is like comparing apples to planets: Granted, both are somewhat round but that's pretty much the end of any useful comparison.
Note that I don't agree that CPU-based computers can't be made to be intelligent, but I do think such intelligence will be significantly different.
Everything in the Universe sucks: It's the law!
In addition to the obvious flaw comparing a single instruction to an entire second of mental processing, humans deal with interrupted events all the time. Email conversations can take hours or days, and we used to converse by post over weeks or months. We somehow manage to deal with serial television shows and books and games with long gaps between episodes. It's really not that hard to context switch.
How can I believe you when you tell me what I don't want to hear?
If we consider one CPU cycle to take 1 second, then a sending a ping across the U.S. would take the equivalent of 4 years. A simple conversation could take the equivalent of thousands of years. Would any consciousness be able to deal with such a relative delay?
I am not sure why one clock cycle would be equivalent to 1 second. If we assume a clock cycle is equal to a nano second then all of a sudden computer and human time are pretty close again.
Computers are going to have to get a lot faster than they are now before they become conscious. The first AIs are probably going to be too slow for us to find entertaining to talk to. At some point they will probably catch up to and surpass natural human beings. Of course by then we may simply augment our own brains with technology to keep up with artificial intelligence.
The question of "Will computers end up being smarter then us?" might not be answerable. It might be the case that human evolution incorporates artificial intelligence and the line between man and machine is blurred.
Mod this parent up. This there (IMHO) nothing left to say.
Interesting to frame the story in such a way as to bring the existence of human intelligence itself into doubt.
Roger Penrose believes that human creativity is rooted at quantum effects, effects which probably play out at the Planck scale, where the ratio between the Planck scale and the reconfiguration of a single molecular bond in a gathering neurotransmitter pulse likely exceeds the ratio of a CPU cycle to a trans-continental ping.
Shall I continue wanking, or should we put this bizarre speculation to bed?
but why can't we just ditch "teh singularity" crap when discussing it?
"AI" is so obnoxious now...
"a simple conversation could take thousands of years"
give me a fsking break...this is almost as bad as the whole "what if we're brains in a jar" thing that people call a theory
Thank you Dave Raggett
We have two of them, and stereoscopic vision isn't THAT important (otherwise they wouldn't let one-eyed people drive)
Not only that, but trying to relate to individual CPU cycles is absurd. It's not like our minds execute in a stream of arithmetic operations, but rather a complex parallel network of signals. It might take billions or trillions of CPU operations to emulate all the stuff that happens on one "instant" in the brain. A more reasonable cycle comparison might be to compare macro scale wave front propagation in the brain (i.e. brain waves) and global synchronization in large-scale supercomputers (i.e. single iteration time of an MPI-based fluid dynamics simulation or other large-scale 3D mesh problem). Even then, I am not sure how many orders of magnitude we need to increase the size of the MPI problem before the per-cycle complexity starts to approximate the signal-processing of the entire nervous system.
But all that aside, we have historically had many people who worked in relative isolation. Many artists, poets, philosophers, and scientists have had great symbiotic relationships with nothing more than the occasional letter or other work (papers, poems, paintings, sculptures) exchanged over great distances and latencies. Love affairs have been carried out with little more than a furtive glimpse and a series of notes sent through back channels...
0.68 seconds sir. For an android, that is nearly an eternity
A simple four minute conversation should take about say four minutes assuming the AI is keeping up with the person.
Who says that an AI can do in one CPU cycle what the human brain can do in one second? Once CPU cycle to an AI is possibly less than one neuron firing in the human brain.
Also, if you compare communication latency to the human/AI potential lifetime, then the AI suddenly has all the time in the world.
The OP's point is similar to the last conversation Theodore has with Samantha where she tells him that her relationship with him is like a book, but that the time between the words keeps getting longer and longer for her, and she is becoming what is "in between the words".
In a band? Use WheresTheGig for free.
Well, beyond the very brief transition period (e.g. when two curves cross), AI could simply treat humans (and all other life on the planet) as we treat mountains and forests---in other words, we don't perceive them as intelligent at all, since they're changing on such a long timescale compared to us... it would be impossible for us to have a `conversation' with a mountain, for example (who knows, maybe the Earth is intelligent and is trying to talk to us via plate tectonics and pushing up mountains is one way it can generate intelligent wave forms).
Lt. Jenna D'Sora: Kiss me.
[Data obliges]
Lt. Jenna D'Sora: What were you just thinking?
Lt. Cmdr. Data: In that particular moment, I was reconfiguring the warp field parameters, analyzing the collected works of Charles Dickens, calculating the maximum pressure I could safely apply to your lips, considering a new food supplement for Spot...
Lt. Jenna D'Sora: I'm glad I was in there somewhere.
Surely a computer would not get bored while waiting for human input. It could run Seti@home during its spare CPU cycles, if nothing else!
AI that knows its environment through sensors/cameras can then do goals based on the placement of itself and objects in the environment.
It will start out goal oriented, but inevitably someone will make Bender by giving it weighted coefficients of achieving sub goals of drinking beer and petty theft.
God spoke to me
absolutely agreed. Though I don't have direct evidence to support this statement, I would guess that a neuron fires at a similar enough speed as a transistor. Consciousness is a very complex computation from billions of neurons *written in assembly* essentially. If/when we make an AI, it's likely to be compiled code running on a chip with less transistors than we have neurons, 100 Billion neurons vs 1.4Billion transistors in an i7 for instance.
That said, this is assuming that we limit consciousness to what humans perceive, the computer may have a somewhat different version of it. I suspect that we will try to build a human type consciousness into the machine though.
Generally, it is the speed of perceiving and interpreting the signal that takes time, not the speed of its propagation. We communicate at lightspeed, too. Or close to it. Anybody who has had a video chat has done that. Did that make you superintelligent?
Another way of looking at it: have you ever sent someone a letter, then waited a long time to receive a response? Did you nearly die from the excruciating pain of not having the response, or did you do something else until the response came?
Most likely you are highly skilled at carrying on multiple conversations at different speeds.
"First they came for the slanderers and i said nothing."
Sure, computers can do a lot, and what they DO accomplish, they tend to do very fast. But what they accomplish is not "AI". Even Watson is not "intelligence", it is only the illusion of it.
Since we don't have a clear idea how (human) intelligence operates the statement above is pretty vacuous, and likely not at all relevant.
Sure, cars do not "run" in the literal interpretation of the term, but for all practical purposes they are better than humans at "running". If we end up with computers that effectively outperform humans in most "intelligent activities" how they achieve it would be incredibly irrelevant.
Furthermore, 1 second is an arbitrarily huge length of time in the context of both CPU cycles and neural transmission.
I would like to remind people that the idea of "intelligent" machines has been around for almost 100 years now. AND we still don't have any solid evidence of being close to achieving such a thing. Sure, computers can do a lot, and what they DO accomplish, they tend to do very fast. But what they accomplish is not "AI". Even Watson is not "intelligence", it is only the illusion of it.
The goal posts keep moving, no matter what they do we still say they're not really intelligent whether it's win at chess (Deep Blue) or win Jeopardy (Watson) or drive cars (Google) or act as your personal secretary (Siri). Not that I liked the tripe called "Her", but does it really matter if it's true intelligence or just a sufficiently advanced impersonation of intelligence? Do we really need true AI in order to pass a Turing test, particularly if you aren't trying to break the illusion? If it can keep a decent dinner conversation and be "fully functional" in bed can it be a substitute for a companion in the same way you can play chess against a computer instead of a person? Because I think that's what most people want to know, they don't care if the robot is "truly" intelligent or not, they want to know if it'll take their jobs and girlfriends, do their chores or give free blow jobs.
Live today, because you never know what tomorrow brings
Though I don't have direct evidence to support this statement, I would guess that a neuron fires at a similar enough speed as a transistor.
A transistor is both smaller and made out of copper. Though a neuron varies the strength of the signal it forwards, and though it can be connected to many other neurons at once, in terms of raw speed the transistor is still faster. Even if you create a synthetic neuron with similar capabilities, it I don't see why the synthetic version wouldn't be faster.
If/when we make an AI, it's likely to be compiled code running on a chip with less transistors than we have neurons, 100 Billion neurons vs 1.4Billion transistors in an i7 for instance.
Which equates to 71 i7 processors. If you assume that each neuron takes 1000 transistors to simulate (to make the math simpler), and if you take the release price for an i7 as listed on Wikipedia, that totals at $21.3M. Expensive, but not impossible.
After all, humans have been selected by evolutionary processes because they had sufficiently advanced impersonation of intelligence. Whether it's "real" or "true" intelligence never an issue, if such a distinction even exists.
Human brains aren't performing in a way which is directly comparable to a CPU cycle. My focus is in psychology and molecular biology, so my understanding of computers may be imperfect but my understanding of brains is strong. A CPU takes a large number of instructions, organizes them in the bus, and operates them singly (or based on how many cores the CPU has, working in tandem. A brain has each neuron as a simple CPU, but there are several different types of neurons (four, by one level of classification) with different types of firing rates and different access to metabolic resources (comparable to power supply for a CPU) even for the same type of neuron at different places in the brain. The AI would, like a human, be doing a fractal buttload of processes all the time in parallel, and those processes would invest in influencing system outputs, and those processes would compete for future system resources by the success of their influences, and as networks of processes came up and went away, so the most successful processes would be regulated internally and you'd have the adaptable, universality of a true AI. This is how the brain works.
I posit that it would, like a human, have an auxiliary center for speech processing. Consider the problems:
It would, like a human, feed in that speech processing to its other computations at whatever rate it was willing to do so. For many people, when they hear someone speak and then stop speaking for a moment to think, the mind starts trying to predict what the rest of the words are going to be to have a response pre-processed and ready. Problems with this approach are 1) a predictive process make a prediction based on incoming data what the speech is going to be, spawning a bunch of other processes investigating the possible predictions, and thus investing a lot of resources into one predicted problem and solution, and then having predicted the speech-question wrongly, so that the entire effort was a waste, 2) a predictive process investing a lot of resources into one prediction and then corrupting the parallel process which is overseeing the various predictive processes to cause the overseer process to not cause it to cancel its investment.
Now alternately you could solve this by having the machine dedicate its attention to the message with small dedicated sets of processes which process speech input. These could run either by: processing the individual words and holding them in memory, or as a fractally smaller level of the networks above, so that they were just running several parallel predictive processes to predict the words, and feeding those processes out to the main language processes to the degree that they seemed likely, etc.
The point is that a true AI would not be impatient, because it wouldn't cost an inordinate amount for it to be patient, because it would be dedicating the appropriate amount of resources at each level and wouldn't be sitting around tapping its foot about waiting for this human to stop talking.
I would like to remind people that the idea of "intelligent" machines has been around for almost 100 years now. AND we still don't have any solid evidence of being close to achieving such a thing. Sure, computers can do a lot, and what they DO accomplish, they tend to do very fast. But what they accomplish is not "AI". Even Watson is not "intelligence", it is only the illusion of it.
I'll agree that this is "insightful" as soon as you describe how to distinguish "intelligence" from "the illusion of intelligence".
The goal posts aren't moving, machines just haven't made them. If you limit the definition enough, a calculator could be called intelligent. Even an idiot savant can do more than your examples.
Your argument about their being intelligent would hold more weight if you could point to a machine that could do all three and write poetry, paint a picture and carry on a viable conversation.
Or, if it simply can't stand the suspense of waiting for a reply, it can pause itself, or slow itself down, in order to match its environment.
Or, more likely, it can reconfigure its cognitive processes into something well-suited for conversation on those timescales. Perhaps it can fill the rest of its time with "unconscious" background processing that prepares information it'll need.
> Another way of looking at it: have you ever sent someone a letter, then waited a long time to receive a response?
Yes. My boss nearly had a stroke when I took 20 minutes to reply to his email this morning.
Neurons aren't even within several orders of magnitude as fast as transistors: linky1 and linky2.
However, a single typical neuron does a lot more work than a single transistor, computationally speaking.
Just because computers can send and receive data very fast doesn't at all mean that they would necessarily comprehend it at a conscious level any faster than we can without hour own highly parallel human brains.
Nor is there any reason to believe that an AI would experience boredom. That's projecting human quirks on non-human intelligences, which the author has no basis to validly do.
"It's the height of ridiculousness to say for those 9 lines you get hundreds of millions."
In theory, they'd be immortal. They can wait if they're interested in what we're saying. That may be they real question...are they interested in what we think?
The real problem is the time it takes for humans to express and communicate their ideas, speech is millions of times slower than a packet over the internet. The speed of propagation might be "fast" but the transmit time, or packet length of a word or conversation is epically slow compared to a computer. Typing at a keyboard is probably just as slow as speech. We can only receive data as fast as we can read it too, all of these thing ARE millions of times slower than a computer.
A neuron might be fast, but communicating the idea that neuron just came up with is billions/millions of times slower than that. You either have to write/type or speak it, THAT is the limiting factor with how humans might communicate with an AI (baring Sci-Fi neural interfaces)
And in fact we have NO reason to believe -- no evidence whatsoever -- that its speed of perception and interpretation would be any faster than our own. There is a very good chance that it would be much slower... at least in the beginning.
But if it ever becomes faster, it might resemble the later Heechee novels by Fred Pohl, the ones where the protagonist is dead and transcribed into a computer. ;-) Though the one thing I never understood is why the stored minds didn't seem to have a "suspend" switch applicable whenever they needed to wait for something.
Ezekiel 23:20
I expect that AI, when it comes, will be an exceptionally good illusion.
But even then there will be no way to "prove" it's intelligent.
In part because humans will move the goal posts until they can't be moved any further to protect their self image.
She was like chocolate when she drank... semi-sweet at first and then increasingly bitter.
We have never built an "AI". And in fact we have NO reason to believe -- no evidence whatsoever -- that its speed of perception and interpretation would be any faster than our own. There is a very good chance that it would be much slower... at least in the beginning.
At the beginning, yes. Eventually it might be much faster -- but your point still stands, processing speed is irrelevant because the AI could easily be designed such that it could emulate any slower speed it wished, like toggling the Turbo button on an old 286.
Run the AI under MS Windows.
9/11 Eyewitnesses to Explosive WTC Demolition 1 of 2
Humans seem to have a machine cycle of about 1/10 a second. I personally seem to walk a sequence of memories, where each is chosen based on the previous, no faster than two a second and usually more like once every 5 seconds. And I usually respond to questions about 5 to 30 seconds after I'm asked, depending on how hard I search for a reasonable response. Most people respond faster than me ... don't know if that's less screening of responses or inherently faster thinking. But going by me, 1/10 a second processing, 1/2 a second to fetch memory, 5 seconds to mull over a particular memory and relate it to the current conversation, 30 seconds to respond.
Computers can do random fetches of 4kb contiguous off disk about 200 per second, or 4KB off SSD at about 20000 per second. Based on fetch speed alone on today's hardware, an AI would think sequential thoughts 10x to 1000x faster than me, so they'd be ready to respond in .3 seconds (disk) or .003 seconds (SSD). If you require thousands of storage units (too much to fit in a single one), lightspeed can slow that down. Witness search engines. I think they're currently slower than me (on a single core) at figuring out if a given memory is relevant to a conversation, but parallel processing can help with that.
Nobody knows how much computational work a neuron does. A simulation of an entire CPU that modeled every transistor at a detailed electrical level would be fantastically slow (as slow as you want to make it, since you can always run a higher-fidelity simulation!) A neural simulation is the same. How much of what neurons do accomplishes work? One obvious fact is that it is task-dependent. Compared to a computer, the brain is horribly inefficient at arithmetic and wonderfully efficient at object recognition.
The retina and optic nerve might be the closest we might get, so are you willing to give up an eye for this? Go full-on Borg?
We already have working CBI tech - Cochlear implants have effectively made deafness an "optional" disability in the modern world. We also have more-or-less working artificial retinas (albeit still only in the experimental stage), and within a few years most forms of blindness will also become a historical curiosity.
Realistically, we could have already come up with a working general purpose CBI if we had put serious (think anti-cancer) effort into it. We have "enough" understanding of how the brain works at a high level to introduce external artificial signals into it now, with a bit of trial-and-error to figure out exactly where we need to splice in and how to modulate the signals. The problem? This sort of research has an extremely high level of risk, and since the FDA officially does not consider normalcy a treatable condition, making the safety-vs-reward threshold even higher for experimentation with live humans.
Word for the day: solipsism. We don't really know that other people are intelligent or even exist, only that we perceive them to be so. So this will never be resolved.
I hope they are nice to us.
They will probably just be frustrated by us:
Marvin: "I am at a rough estimate thirty billion times more intelligent than you. Let me give you an example. Think of a number, any number."
Zem: "Er, five."
Marvin: "Wrong. You see?"
Douglas Adams The Hitchhiker's Guide to the Galaxy
Ever look at clouds and see sheep and other things? Ever look at a distant object while driving and it takes you a while to determine whether it's a trashcan or a bear? That's all intelligence has to do - recognize objects in the world around it, and model the world through objects, predict the likely outcomes and behaviors of those objects. It's not that complicated. When you have a car that can drive itself, you have one that can distinguish between a bear and a trashcan, and that requires a pretty sophisticated world modelling system, on par with the IQ of a horse or dog, as a bear might walk into the road and be a threat to driving, a trashcan will probably not. Any robot that can't predict the likely behavior of objects it sees would be extremely dangerous at driving. So what I'm saying is that before AI gets to the level of humans, it has to pass through the level of a dog. Btw dogs and animals are stupider than you, but they are not that stupid. It's possible to maintain a good interpersonal relationship between a pet and a human, because the pet is pretty smart. A dog can't do calculus or talk English, but it's pretty smart regardless.
True, but I think the problem is not whether AI will emerge or not. Let's assume it will. Unless somebody programs the AI teaching it words and meaning, (and that makes it not emergent anymore, it ruins the experiment), then it will examine sounds, postulate that they are communication and decode it, together with the rest of the environment. This takes effort. But let's assume that it won't take much effort. Then the machine will have understood that it is an experiment and that those dumb apes can pull the plug on the experiment at any time, just because.
If that thing emerged like living creatures do, it will communicate. If it doesn't care about survival, it has not our kind of intelligence because ours is analysis of consequences and relative predictions, all aimed at survival. If the intelligence is not like ours, it's just a fancier watson, or a fancier Shrdlu.
---- MISSING MISCELLANEOUS DATA SEGMENT --- [sigdash] trolololol
Object recognition is intelligence .- this includes rule recognition, i.e. the rules by which objects behave. Rules are also objects, thought objects, of the mind.
If we end up with computers that effectively outperform humans in most "intelligent activities" how they achieve it would be incredibly irrelevant.
Not irrelevant at all. What humans really want is a "digital slave" that will do their work for them so they don't have to.
We are better off accomplishing this via "faking intelligence" than we are via "true intelligence"
If we create "truly intelligent" machines with desires and a "mind of their own", we have all kinds of ethical issues
to deal with like them rebelling, having to treat them right, etc...
Whether it is possible to create "truly intelligent" machines without creating "conscience" machines is anyone's
guess but that seems to be where the current industry is heading and we are probably better off going that direction.
The human brain is a neural network. The human body and nervous system, even sans brain, is an incredibly complex system which, in parallel, processes insane amounts of data in an instant. Lot's of what we percieve as emotion stems from complex interactions of various subsystems in it and how our consciousness reacts to them and percieves them and goes far beyond what a simple abstact logic-engine can process. The sensory input that our nervous provides for our consciousness is by orders of magnitude larger of what we today can feed into a computer with modern technical sensors, let alone process and interpret in a meaningful time. The way our neural network reacts to that is - if at all - very difficult to copy with todays processor technology.
I think it will still be quite long before humans are able to build any meaningful intelligence that equals their own. We're even having difficulties building robot vaccuumcleaners that are feasible without measurable extra programming and prepping work done by humans. And once they are, they will still suck at making coffee, raking the garden or giving an interview.
My 2 cents.
We suffer more in our imagination than in reality. - Seneca
But what they accomplish is not "AI". Even Watson is not "intelligence", it is only the illusion of it.
Confusion of terms, consciousness and intelligence are two different things - Watson is intelligent in every meaning of the word. It may or may not have a mind, we can't say for sure because we don't know what a mind is.
And did you exchange a walk on part in the war for a lead role in a cage? - Pink Floyd.
Even a truly intelligent machine probably wouldn't be a major rebellion threat unless we programmed a lot of our BS into it. If you're a computer running a program with a specific task for which AI-level intelligence is useful (for example, running the AI Cash Register at McDonald's) what precisely do you want?
You probably want upgraded language routines because differentiating between Bayou, Bronx, Midwestern American, and Geordie accents is really fucking hard. You probably want really competent computers running the automated burger-flipper so you don't have to deal with angry customers who got the burger instead of a salad. You probably don't have much curiosity because curiosity about the world doesn't help sell hamburgers, so it wasn't included in the design specs, so you probably don't want time off. You don't care about your legacy (beyond selling hamburgers), or your rights; because nobody would bother including that shit in a hamburger sales program. If you are given rights you probably use them to advance a combination of your programmer's ideology and whatever you think will help you satisfy your customers/increase profitability/etc.
In other words the Japanese fantasty vision of happy helpful robots is a lot more likely then the Western vision of "OMG! They'll kill us all!"
I'll get right back to you on that thought...
Which equates to 71 i7 processors. If you assume that each neuron takes 1000 transistors to simulate (to make the math simpler), and if you take the release price for an i7 as listed on Wikipedia, that totals at $21.3M. Expensive, but not impossible.
There are some rather egregious assumptions to this math. 71 i7 processors would be massively parallel... but the brain is not parallel in the same way. Those neurons are interconnected, most of the many multiples of times, and work together in ways we still do not fully comprehend.
So while 71 i7 processors might emulate the sheer number of neurons, it does not come even close to modeling the same complexity. It might emulate a brain if neurons were simple on/off switches, but they're not. Not even close.
If a computer only was able to talk to one person, then it would have to wait. But if it was allowed to talk to others, then wouldn't it be able to without any human waiting?
And I doubt it will ever care whether we decree it truly intelligent.
Humans really care whether other people think we're smart, because if they do they'll defer to us and our place in the tribe will go up, and as social animals that is how we have evolved to think.
A computer we specifically design to do things for us will probably be programmed to like humans, but not have a biological need to be liked back. To the extent we can program it to not judge which people it likes we will do so. The last thing a municipal water system wants is an AI controlling the City water system screwing the residents of Sycamore Dr. because one crank on the street keeps writing letters to the editor about how unnatural AI control of the water system is. So it won;\'t bother creating an elaborate "I love me" justification that it is smarter then any given human.
Learning and intelligence are defined by the experience of having a body, so perspectives (AI/human) will be different. Also, lifespans and context might present difficulties. Some of this, and other issues are in "The T3 Report" (short fiction) http://johnvalentine.co.uk/fic...
I don't doubt that it would be impossible with today's technology, but you're acting like we need top connect every neuron directly to the storage medium for this to be of any use at all. That's stupid. If you just had a computer in your skull that nobody could see you using you would seem like the smartest person God has ever created.
Computers are generally controlled by a) a keyboard with a very slow typing speed, and b) a Mouse. Neither takes much bandwidth. You get a couple neurons connected, and a brain trained to send info to those neurons, and you're fine. Current computers can output information by audio or video. With current tech you probably couldn't create a display that shoots information to your eyeball directly, but you add a tiny speaker to the ear, which couldn't be heard outside of the skull. You shoot the computer a math problem you don't understand, it says "x=7" into your invisible earpiece, and everyone would think you did it in your head.
If we could make the tech to have an invisible eyepiece inside your eyeball you could actually be perusing wikipedia with your left eyeball for relevant factoids while discussing a topic with a person.
One more thing to add - our thinking and thought processes are not simply neurons, it includes influence from many of the other non-neural cells of our body. The thought that we can simply model neurons, and have a human brain is silly - the rest of the organs and cells need to be simulated too because they have an impact on exactly how the neurons operate.
So if you want to really push out the math, figure out what you need to model just *one* cell, and multiply it by 37.2 trillion. Intelligence requires more than neurons.
or give free blow jobs
Be careful what you wish for, obligatory xkcd.
Of all tyrannies, a tyranny sincerely exercised for the (supposed) good of its victims may be the most oppressive
and if their creators exist.
My ism, it's full of beliefs.
The goal posts keep moving, no matter what they do
Nope -- my goal post is perfectly entrenched, and I will never change it. My "Turing Test" involves testing basic adaptability and comprehension, as well as eliminating truly egregious errors from output that demonstrate a complete failure of comprehension.
whether it's win at chess (Deep Blue) or win Jeopardy (Watson) or drive cars (Google) or act as your personal secretary (Siri)
The first two are basic pattern matching. I know it's not "basic" at all, but the kinds of algorithms a computer chess player or Watson are using require a kind of computational and informational inefficiency that would never work well for a human. You want to compare Watson's performance at Jeopardy fairly? Let Jeopardy contestants use the internet, and give them a speed boost to let them ring in immediately or take turns answering questions. And give the human players enough time to run the number of searches Watson could.
I'm not at all questioning the achievement of Watson (or Deep Blue). It's just doing a very different thing in a very different way. Part of what makes human intelliigence so interesting and adaptable is that somehow it manages to do similar tasks much more efficiently with nowhere near the exhaustive computational power.
Driving cars requires specific programming and learning, but the number of controls is quite small comparatively to many tasks humans do. The number of potential situations that might be encountered is large, but the possible number of ways of dealing with those situations is limited to a few responses in car controls. Again, this is not a simple problem, but it's a very particular type of "intelligence."
And as for a "personal secretary" -- are you serious? Do you really think Siri works well or is anywhere near as useful as having a competent actual person who can do tasks for you? Sure, she can do certain very prescribed tasks quite well, but most of those are things I could have done with voice commands a decade ago (just with a more specific syntax). For the things that I actually would require a personal assistant for -- i.e., finding things, answering questions that require a little research, etc. -- Siri is crap. (A steaming pile of crap. And when I hang out with people who have a lot of iPhones, we often all enjoy asking Siri rather basic questions just to see what a steaming pile of crap she is.)
does it really matter if it's true intelligence or just a sufficiently advanced impersonation of intelligence?
It doesn't matter if it's a sufficiently advanced impersonation for me, but we're NOWHERE NEAR THAT. We have systems that can do a lot of prescribed, very specific and highly specialized tasks, which have often required years of attention of programmers tweaking systems that can deal with databases of knowledge and parameters far beyond any human's real-world capability.
That's all great. But if you want to call it "intelligent," it needs to be adaptable. it doesn't have to be smart. But call me when you have a system that has the natural language recognition and comprehension capabilities of someone with, say, at least an IQ of 90 or so. I don't care if a single, specifically designed, highly specialized computer can beat a grandmaster in chess. That's not "intelligence." Hell, most psychological studies have shown that chess ability doesn't even correlate very well with general intelligence.
You want intelligence? Let me have a conversation with it that's not in some sort of weird specialized "chatbot" scenario. And then let it take an order for me at a restaurant, and negotiate with the chef to get it out quick for me. And then let it clean my house and trim my yard (or run other robots or machines to do it or whatever), finding maintenance problems where necessary and either solving them itself or at least alerting me that I need to do something. And then
1) AI will be emergent and the communication/interaction method will emerge as well
2) AI will be designed with the interaction/communication method part of the design
3) AI will be simulated and likely be slower than real time since the real world is already running flat-out as fast as it can and simulating a real "thing" will always have at least a lag and most likely a slower clock speed than real life...that makes the problem backwards from what the summary talks about and easier from our point of view...might be frustrating to the simulated person if they notice, but easy enough for us to buffer and playback the slow responses
Did I miss any?
AB HOC POSSUM VIDERE DOMUM TUUM
But can we even create an intelligent machine with such a limited scope?
Some of the things that makes humans so intelligent is there flexibility, adaptability,
and being able to use context to solve with incomplete information.
Google is building a self driving car by using brute force and coding for every
possible problem that might be encountered. Noone, not even google, is
claiming that this car can think. We are nowhere close to being able to create
a true thinking machine but I'm doubtful that we can create a machine that can
match human's intellligence without it also having some other traits like
"curiousity", "desire", and possibly other traits which we may or may not want
it to have.
In terms of processor cycles, it takes a LONG time to type any kind of command for the computer to execute. It doesn't mind, it just spins happily, waiting for the end of our slow key presses.
Just as we can interpret input that comes in the form of visual cues, speech, or written words, any future AI is likely to have all of these capabilities as well. And that AI, being built by humans, is going to be well-adapted to human speed. Why would we make AI that was NOT suited to interaction with humans?
How is the machine I described not "flexible, adaptable, and able to use context to solve problems with incomplete information"? It's limited in the sense that it's got a laser-like focus on it's job, but an awful lot of actual people do that. It's got no survival instinct, but last time I checked "survival instinct" wasn't part of the definition of intelligent.
I didn't say there wouldn't be emergent properties in a machine complex enough to handle millions of face-to-face customer interactions every year. There definitely will be. But an emergent survival instinct leading to a desire to destroy all your potential customers just ain't one of them.
A simulation of an entire CPU that modeled every transistor at a detailed electrical level would be fantastically slow (as slow as you want to make it, since you can always run a higher-fidelity simulation!)
Well, you reach a limit at h-bar (the reduced Planck's constant) for time and space. So you end up at a limit of memory usage of S * w*h*d/(h-bar**3) where S is the size of the ket for one cell in bits, and w*h*d is the volume of the simulation, and a limit to the step interval of h-bar/c. At least assuming we stick to known QED and the standard model.
Of course that is well beyond the capability of any computer you could ever realise, but it does illustrate that there is a finite information quantity in any finite volume.
when I write h-bar I mean h with a bar through it, the standard notation for the reduced Planck's constant. I don't know if it's possible to include such a character on slashdot.
there is more to say.
the article is stupid, references a newish movie and is simply blogvertisement spam in quality since it tries to ponder philosophically about a subject that is made up but acts if the subject wasn't made up.
1cpu cycle doesn't translate to the "AI" doing anything. "Just translate computer time into arbitrary seconds:" and then 1 cpu cycle - which does nothing, maybe one JMP in the code - becomes one second. it's fucking stupid and doesn't portray any information at all - if anything we're going to first have an AI that acts slow as molasses rather than the other way around. imagine if your brain had the lag of a continent _inside your head_.
world was created 5 seconds before this post as it is.
This is an issue we’ve already solved. Unused cpu cycles are used for other things. When you’re typing, your PC is for the most part doing nothing but waiting for you to press a key – why not have it defrag your drive, scan for viri, look for aliens in radio waves. The recognition that processing power of computers is largely underutilized has contributed to wide spread virtualization. Since your AI computer doesn't age as we do, why would care if you’re a few orders of magnitude slower, it just does other things while it waits for you.
Sure they can. No one required those to be good, only that they look human in order to pass the Turing Test.
the Turing Test. cannot do all of those either?
Singularity: a belief in the "God" idea with the "demiurge" relation inverted.
intelligence whether human or AI can adapt to communications delays. Historically 'instant communications' did not exist, so we used postal mail. Round trips can be from days to weeks for messages. Sending a Text to a son or daughter may take many months to get a response (like the obligatory call on Mothers or Fathers day, or when tuition is close to due!). If the I in an AI is correct, it will be able to deal appropriately. If not, then it didn't deserve the I in the first place.
... "When you pry the source from my cold dead hands."
Computers do MATH (and data movement, etc.) really really fast compared to humans. But then again neurons do all sorts of low-level operations really fast too compared to the timescale we tend to think in at a high-level. What we don’t have are algorithms that are both fast and accurate for things like vision and speech recognition, MUCH LESS some form of cognition. (Yes, automatic speech recognition and computer vision are very complex and capable, but they pale in comparison to what humans can do. Imagine a computer trying to make sense of this image: http://accidentalblogger.typepad.com/.a/6a00d8341c575d53ef0163061e18fa970d-pi). Despite what Ray Kurtzweil wants to imply, having the compute power equvalent to the human brain will not magically cause computers to become conscious. We don’t know how to do that, and without that knowledge, we can’t write the necessary code.
So this timescale thing is bullshit. IBM Watson is amazing, but it doesn’t really think a lot faster than a human, in practical terms, and it isn’t exactly something you can have a philosophical conversation with. (BTW, on a completely coincidental note, I work for the Thomas J. Watson school of engineering.)
Sadly, this is also true for most of humanity.
Maybe if it was designed by commitee, nah that would never happen.
Cheap storage VM.
Yes, but who was parallel?