Slashdot Mirror
Why Robots Will Not Be Smarter Than Humans By 2029
Hallie Siegel writes "Robotics expert Alan Winfield offers a sobering counterpoint to Ray Kurzweil's recent claim that 2029 will be the year that robots will surpass humans. From the article: 'It’s not just that building robots as smart as humans is a very hard problem. We have only recently started to understand how hard it is well enough to know that whole new theories ... will be needed, as well as new engineering paradigms. Even if we had solved these problems and a present day Noonian Soong had already built a robot with the potential for human equivalent intelligence – it still might not have enough time to develop adult-equivalent intelligence by 2029'"
Kurzweil's predictive powers are so incredibly wrong that he could literally destroy the world by making a mundane prediction that then couldn't come true.
For example, if Kurzweil foolishly predicted that the sun would come up tomorrow, the earth would probably careen right out of its orbit.
AntiFA: An abbreviation for Anti First Amendment.
By the same argument you could say that any good library from 1950 was also smarter then a human. You'd be just as wrong.
John McAfee 'It was like that time I hired that Bangkok prostitute; to do my taxes, while I fucked my accountant'
Robotics expert Alan Winfield offers a sobering counterpoint to Ray Kurzweil ...
I like how the naysayers are depicted as sober, rational minded individuals while those who see things progressing more rapidly are shown as crazy lunatics. They are both making predictions about the future. Why is one claim more valid than the other? We're talking fifteen years into the future here. Do you think that the persons/people predicting that "heavier than air flying machines are impossible" only eight years before the fact were also the sober ones?
Lord Kelvin was a sober, rational minded individual. He was also wrong.
Look where all this talking got us, baby.
Analysis: By 2029 people will be so dumb that current robots will be smarter than humans.
Computers on the other hand can already be argued to be smarter than a human - if you consider the entire internet as a single computer.
Depends on how you define "smarter."
The internet holds more knowledge than a single human ever could, but machines cannot do anything without direct, explicit directions - told to it by a human. That's the definition of stupid to me: unable to do a thing without having to all spelled out to you.
There's a reason D&D considers Wisdom and Intelligence to be separate attributes.
An enigma, wrapped in a riddle, shrouded in bacon and cheese
In a large number of ways, a 1950's library is smarter than any human.
If the measure of "smart" is how closely it behaves like a human - sure, we're probably a ways off. ...we're making progress, but have a way to go.
If the measure of "smart" is what we know (in bulk), we're already there.
If the measure of "smart" is the ability to synthesize what we know in useful relevant ways...
Oblig. xkcd: https://what-if.xkcd.com/5/
What a fool believes, he sees, no wise man has the power to reason away.
If the contents of my Facebook feed can be taken into consideration, one could reasonably make the argument that robots are smarter than humans right now.
We're going down, in a spiral to the ground
AI suffers from continuously moving goal posts because nobody has a good definition of intelligence.. A computer (Watson) has already convincingly beat humans at general knowledge. Watson is an amazing technological feat however the general public does not recognise Watson as intelligent in any meaningful way, they have the same reaction as my wife when they see Watson playing Jeopardy - "It's looking up the answers on the internet, so what?". They don't even understand the problem Watson has solved, when the general public talk about AI they are thinking about robots that appear in modern movies and are basically indistinguishable from humans (eg:Terminator), something that is not only intelligent but also has also (nearly) mastered human social intelligence.
In a way they are right, emotions drive what the logical mind thinks about and AI cannot (yet) communicate, let alone reproduce, human emotions, I have long thought that this is partly because AI researchers in general concentrate on modelling the brain and more or less ignore the huge network of intricate sensors and actuators attached to it.
And did you exchange a walk on part in the war for a lead role in a cage? - Pink Floyd.
I certainly wouldn't argue that libraries are self-aware.
It all goes back to the definition of smart is. Libraries certainly contain more information -- at least, in a classical sense. [Maybe one good memory of a sunset contains more information - wtfk] Watson, for example, is just a library with a natural language interface at the door. By at least one measure -- Jeopardy :) -- a library (Watson) is smarter than a lot of people.
will be what causes the singularity!
o we don't know what "thinking" is -- at all -- not even vaguely. Or consciousness.
o so we don't know how "hard" these things are
o and we don't know if we'll need new theories
o and we don't know if we'll need new engineering paradigms
o so Alan Winfield is simply hand-waving
o all we actually know is that we've not yet figured it out, or, if someone has, they're not talking about it
o at this point, the truth is that all bets are off and any road may potentially, eventually, lead to AI.
Just as a cautionary tale, recall (or look up) the paper written by Minsky on perceptrons (simple models of neurons and in groups, neural networks.) Regarded as authoritative at the time, his paper put forth the idea that perceptrons had very specific limits, and were pretty much a dead end. He was completely, totally, wrong in his conclusion. This was, essentially, because he failed to consider what they could do when layered. Which is a lot more than he laid out. His work set NN research back quite a bit because it was taken as authoritative, when it was actually short-sighted and misleading.
What we actually know about something is only clear once the dust settles and we --- wait for it --- actually know about it. Right now, we hardly know a thing. So when someone starts pontificating about dates and limits and what "doesn't work" or "does work", just laugh and tell 'em to come back when they've got actual results. This is highly distinct from statements like "I've got an idea I think may have potential", which are interesting and wholly appropriate at this juncture.
I've fallen off your lawn, and I can't get up.
To be fair, your ability to tell if the grass needs cut is also based on sampling grass growing patterns over your entire life...
We're probably more than 15 years from strong AI. Having been in the field, I've been hearing "strong AI Real Soon Now" for 30 years. Robotic common sense reasoning still sucks, unstructured manipulation still sucks, and even Boston Dynamics' robots are klutzier than they should be for what's been spent on them.
On the other hand, robots and computers being able to do 50% of the remaining jobs in 15 years looks within reach. Being able to do it cost-effectively may be a problem, but useful robots are coming down to the price range of cars, at which point they easily compete with humans on price.
Once we start to have a lot of semi-dumb semi-autonomous robots in wide use, we may see "common sense" fractured into a lot of small, solveable problems. I used to say in the 1990s that a big part of life is simply moving around without falling down and not bumping into stuff, so solve that first. Robots have almost achieved that. Next, we need to solve basic unstructured manipulation. Special cases like towel-folding are still PhD-level problems. Most of the manipulation tasks in the DARPA Robotics Challenge were done by teleoperation.
That presumption seems to be precipitated on the theory that a computer intelligence won't "grow" or "learn" any faster than a human. Once the essential algorithms are developed and the AI is turned loose to teach itself from internet resources, I expect it's actual growth rate will be near exponential until it's absorbed everything it can from our current body of knowledge and has to start theorizing and inferring new facts from what it's learned.
Not that I expect such a level of AI anytime in the near future. But when it does happen, I'm pretty sure it's going to grow at a rate that goes far beyond anything a mere human could do. For one thing, such a system would be highly parallel and likely to "read" multiple streams of web data at the same time, where a human can only consume one thread of information at a time (and not all that well, to boot.) Where we might bookmark a link to read later, an AI would be able to spin another thread to read that link immediately, provided it has the compute capacity available.
The key, I think, is going to be in the development of the parallel processing languages that will evolve to serve our need to program systems that have ever more cores available. Our current single-threaded paradigms and manual threading approaches are far too limiting for the systems of the future.
I do not fail; I succeed at finding out what does not work.
It has nothing to do with processing speed, or parallel processing. Brains in general, human brains included, do not process information. They generate consciousness. They do this in ways that neuroscientists still don't understand. As a neuroscientist I can say this without hesitation. Silicon chips are not alive, and will never generate consciousness as we now understand it. But they can process information much faster than the human brain.
A brain is a terrible thing to waste... Mind? That's debatable.
Working memory is the space that you actively think on. It's not clear how it's stored, but it's clear that most memory is not just words. An AI will start with an in memory way of storing connected concepts; actors, linguistic, mathematical, logical, not understood but remembered cause/effect, image. Parsing the information into working memory involves putting it into a form that the intelligence can use.
This is a pretty well understood concept. The details are the tricky part.
John McAfee 'It was like that time I hired that Bangkok prostitute; to do my taxes, while I fucked my accountant'
The internet holds more knowledge than a single human ever could, but machines cannot do anything without direct, explicit directions - told to it by a human.
I'm sure not doing anything would still be way better than someone only checking facebook for a whole day. Which increases the score on the Robot side.
Brains in general, human brains included, do not process information. They generate consciousness.
Brains *do* process information. They *also* generate consciousness. I would argue that they generate consciousness *by* processing information.
They do this in ways that neuroscientists still don't understand. As a neuroscientist I can say this without hesitation.
We don't understand how consciousness is generated. That doesn't mean we can't make it happen. The Wright brothers made a working airplane before everything was figured about about aerodynamics. Many aeodynamic principles were at play in the Wright Flyer that the Wright brothers didn't understand, but they knew enough to make it happen.
Maybe humans don't know how consciousness works. But we know how evolution works. And evolution generated consciousness. We may not be able to make consciousness directly, but we may be able to make something that can make consciousness in a way we don't understand.
Silicon chips are not alive, and will never generate consciousness as we now understand it.
In order to make something conscious, the parts need to be "alive"?! Well we know that's not true. Humans are conscious. They are made of cells. The cells are made of proteins that are not alive. Humans are ultimately made of quarks and leptons. None of which are alive or conscious. Clearly living things and consciousness can be made from parts that are not themselves living or conscious.
Siegel is of course right because he can predict the effect of unexpected future inventions, and Kurzweil cannot. Oh wait...
I'm just "this guy", you know?
Terrifyingly, "The Hate" might be one of the easier first things to simulate in AI!
The reason is that it's often demonstrated with a far lower level "skillset" than the smart comments.
See for example the (thinning?) pure troll posts here. Despite the rise in lots of other things, I'm noticing fewer pure troll posts of the worst vicious kind. I wondered idly why they got here so regularly. Anyone remember the ones that went:
"so you sukerz ya haterz loosers you take it and shove it?"
Any 1000 of you could write a 100 line program that can run circles around that!
I still do one day wish to work with any Chattterbot programmer who wants to try some custom mods.
My first Journal Entry ever, in 8 years! http://slashdot.org/journal/365947/aphelion-scifi-fantasy-horror-poetry-webzine
No.
First we need to define consciousness.
Then we get to decide if something fits the definition or not.
I really do not understand why you are acting as if you are unable to grasp that point. Is this some sort of debating trick?
Kurzweil's smart machine predictions are, last I checked anyway, based on a rather brute force approach to machine intelligence. We completely understand the basic structure of the brain, as a very slow, massively parallel analog computer. We understand less about the mind, which is this great program that runs on the brain's hardware, and manages to simulate a reasonably fast linear computing engine. There is work being done on this that's fairly interesting but not yet applied to machine mind building.
So, one way to just get there anyway is basically what Kurzweil's suggesting. Since we understand the basic structure of the brain itself, at some point we'll have our man made computers, extremely fast, somewhat parallel digital computers, able to run a full speed simulation of the actual engine of the brain. The mind, the brain's own software, would be able to run on that engine. Maybe we don't figure that part out for awhile, or maybe it's an emergent property of the right brain simulation.
Naturally, the first machines that get big enough to do this won't fit on a robot... that's why something like Skynet makes sense in the doomsday scenario. Google already built Skynet, now they're building that robot army, kind of interesting. The actual thinking part is ultimately "just a simple matter of software". Maybe we never figure out that mind part, maybe we do. The cool thing is that, once the machine brain gets to human level, it'll be a matter of a really short time before it gets much, much better. After all, while the human brain simulation is the tricky part, all the regular computer bits still work. So that neural net simulation will be able to interface to the perfect memory of the underlying computing platform, and all that this kind of computation does well. It will be able to replace some of the brute force brain computing functions with much faster heuristics that do the same job. It'll be able to improve its own means of thinking pretty quickly, to the point that the revised artificial mind will run on lesser hardware. And it well be that there are years or decades between matching the neural compute capacity of the human mind and successfully building the code for such a mind. So that first sentient program could conceivably improve itself to run everywhere.
Possibly frightening, which I think is one reason people like to say it'll never happen, even knowing that just about every other prediction about computing growth didn't just happen, but was usually so conservative it missed reality by lightyears. And hopefully, unlike all the doomsday scenarios that make fun summer blockbusters, we'll at least not forget the one critical thing: these machines still need an off switch/plug to manually pull. It always seems in the fiction, we decide just before the machines go sentient and decide we're a virus or whatever, that the off switch didn't needed anymore.
-Dave Haynie