Stop Fearing Artificial Intelligence • Hacker News Board

Re: Stop Fearing Artificial Intelligence

#61

Post by neonbat » Sun, Apr 19, 2015, 7:25 PM UTC

icebraining wrote: post #42

neonbat wrote: post #16

Earlier quoted context omitted.

I agree; that is part of his argument. Obviously though it is a completely flawed argument. Up until the creation of the personal computer most machines were specialists. They were large computers designed for specific computation tasks or machines made specifically for manufacturing certain kinds of goods. Generalist computers didn't have much precedent. That's kind of the point of new technology. It's not necessari…

Up until the creation of the personal computer most machines were specialists. They were large computers designed for specific computation tasks or machines made specifically for manufacturing certain kinds of goods. Sorry, but that's just not true. Back in the 60s, way before personal computers, the were many time-sharing companies, selling general "computing resources" to different people who would dial-up into the…

Yeah I thought about that when I was writing the reply. I don't really consider that a generalized use case. Is it more generalized? Yes. Computing itself, however, can be argued to be a specialized task. My computer today doesn't do "computing" it lets me read news stories, it lets me talk on skype, it lets me send mail, it lets me have philosophical discussions on hackernews. The personal computer is so clearly a generalized machine whereas those time sharing rigs back in the 60s were clearly not. They were specialized tools for programmers.

Re: Stop Fearing Artificial Intelligence

#62

Post by quanticle » Sun, Apr 19, 2015, 7:33 PM UTC

david-given wrote: post #46

Netcob wrote: post #24

Earlier quoted context omitted.

The problem with that is that having proof and getting overpowered by a superhuman AI might fall within a very short timeframe. The risk (impact*probability) is simply too high to be ignored. We don't know the probability very well, but considering the impact, it doesn't matter that much. I think if the risk is high enough, the burden of proof gets shifted. To just swat it away with a sentence like "Sounds reasonable…

The problem with that is that having proof and getting overpowered by a superhuman AI might fall within a very short timeframe. To be brutally honest, I don't actually see this as being necessarily a bad thing --- it's not like humans are doing such a great job of running the world. If a hypothetical AI is so much better at it than we are that it can replace us, then don't we deserve to be replaced? Although I would…

It depends on how you define "better". As a sibling comment points out: let's say that the AI is programmed to maximize the number of paperclips in the universe. And then to accomplish that goal, it kills all humans (by eating our infrastructure to turn it into paperclips). But that's not enough. The earth's mantle and core contain massive amounts of iron and aluminum that could be used to make paperclips. So the AI starts consuming the Earth. And then it scans the rest of the solar system... ah, Mars and the asteroid belt contain lots of iron and aluminum too!

I fail to see how a solar system turned into a ring of paperclips orbiting a sun is in any way "better" than what we have right now.

Re: Stop Fearing Artificial Intelligence

#63

Post by foobarqux » Sun, Apr 19, 2015, 7:34 PM UTC

rwallace wrote: post #38

foobarqux wrote: post #21

It becomes much easier to understand the AGI threat if you consider whole brain emulations -- human minds running at superhuman speed on silicon. Hollywood has made it obvious, if it wasn't already, the serious implications of that type of development. Whole brain emulations also make it clear that the issue has nothing at all to do with human values: a human superintelligence would be a catastrophic risk to humanity…

Actually, a whole brain emulation would, at least for some generations after the first working model, run slower, not faster, than an organic brain. Bear in mind that without whole brain emulation (or nanomedicine sufficiently advanced to confer biological immortality, which is an even considerably more difficult engineering challenge), you're going to die. So are all your family and friends. Everyone you ever cared…

The point was that everyone seems to have trouble understanding why software AGI might be dangerous but a much easier time why whole brain emulation is, not that Hollywood is a representation of reality.

Re: Stop Fearing Artificial Intelligence

#64

Post by icebraining » Sun, Apr 19, 2015, 7:35 PM UTC

ikeboy wrote: post #17

chubot wrote: post #12

I read "Superintelligence" by Nick Bostrom, essentially on the recommendation of Elon Musk (he tweeted about it). It talks about the dangers of strong AI and possible paths to it, and how humans can mitigate its effects. The only reason I read past the beginning is because in the preface he says: "This book is likely to be seriously mistaken in a number of ways". So at least he's intellectually honest. I believe he's…

Why do you feel we will not build "Strong" AI? We know AI as smart as humans can theoretically exist, because humans exist, so intelligence is possible. Is there a reason to think that humans are near the upper limit of what's possible? If not, why can't we build something significantly smarter than us?

Is there a reason to think that humans are near the upper limit of what's possible? If not, why can't we build something significantly smarter than us?

A simple hypothesis: because we can't build what we can't understand, and we can't understand an intelligence smarter than us.

(Most theories about the Singularity have machines building increasingly smarter machines, in a recursive fashion, but they could likewise be hampered by the same problem.)

I don't think this is necessarily true, but it makes sense to me, and doesn't require that we are the theoretically upper limit, just the local maximum.

Re: Stop Fearing Artificial Intelligence

#65

Post by icebraining » Sun, Apr 19, 2015, 7:37 PM UTC

ikeboy wrote: post #41

chubot wrote: post #33

Earlier quoted context omitted.

Mainly because I haven't seen any evidence. Special purpose AI is not evidence. That point is debatable, but most AI practitioners agree; it's only some technologists like Elon Musk et. al. that seem to be extrapolating. Sure, humans exist, and we can already make humans by combining DNA from a man and a woman :) That method does not lead to superhuman intelligence. What makes you think we could devise some other met…

>Computers and humans are completely different things. If you watch a child grow up, this is obvious. A child can do things that are impossible for a computer, while it can't speak, read, or write. A computer can do things that no human could ever do. They are qualitatively different. Do you think humans can do things that no computer will ever do? If yes, could you give an example? If not, doesn't that imply that hu…

Do you think humans can do things that no computer will ever do? If yes, could you give an example? If not, doesn't that imply that human-level AI can exist?

Exist, yes. Be built by us, not necessarily.

Re: Stop Fearing Artificial Intelligence

#66

Post by ikeboy » Sun, Apr 19, 2015, 7:39 PM UTC

icebraining wrote: post #64

ikeboy wrote: post #17

Earlier quoted context omitted.

Why do you feel we will not build "Strong" AI? We know AI as smart as humans can theoretically exist, because humans exist, so intelligence is possible. Is there a reason to think that humans are near the upper limit of what's possible? If not, why can't we build something significantly smarter than us?

Is there a reason to think that humans are near the upper limit of what's possible? If not, why can't we build something significantly smarter than us? A simple hypothesis: because we can't build what we can't understand, and we can't understand an intelligence smarter than us. (Most theories about the Singularity have machines building increasingly smarter machines, in a recursive fashion, but they could likewise be…

In certain domains, we've built things smarter than us. Chess comes to mind.

Wouldn't your hypothesis imply that no computer could beat its maker at chess?

Re: Stop Fearing Artificial Intelligence

#67

Post by Rangi42 » Sun, Apr 19, 2015, 7:41 PM UTC

JimDabell wrote: post #44

Moshe_Silnorin wrote: post #4

Nobody is afraid of today's AI algorithms. But if we make machines that are smarter than us and have desires, they will influence the future to achieve their desires. If these desires conflict with our own, things will not end well for the dumber party. As we really have no idea what we, collectively, think of as a moral terminal goal, and less so how to formalize this, there is no reason to expect the first AIs to h…

> if we make machines that are smarter than us and have desires, they will influence the future to achieve their desires. If these desires conflict with our own Aren't you anthropomorphising them here? Just because something might be intelligent, it doesn't follow that they automatically gain other attributes that we have, such as desires. Why would any AI have anything that could meaningfully be called "desires", ex…

Making something intelligent does not automatically give it desires. But as you said, there is every reason to expect that we would build AI to achieve our goals, which means making it (a) desire to achieve a goal we share and (b) intelligent enough to succeed. (This is not anthropomorphizing AI; there's no need to make it feel desire the way we do, it just needs to tend to reach a state where the goal is achieved.)

The danger comes from our inability to accurately define our own desires. There is no doWhatIMean() function. A programmed goal can subtly differ from the one we want so that it passes the obvious test cases but breaks catastrophically in an unusual real-world situation. There's an anecdote[1] about a neural network trained to detect camouflaged tanks (I've also heard of it detecting Soviet versus NATO tanks); it ends up learning to detect objects with shadows, because the tanks were photographed on a sunny day and the non-tanks on an overcast day. Now imagine if the program had been prematurely instructed to attack the camouflaged tanks. You can call this "broken and useless software," but by the time we realize how it's broken it could be too late.

[1]: http://www.j-paine.org/dobbs/neural_net_urban_legends.html

Re: Stop Fearing Artificial Intelligence

#68

Post by troels » Sun, Apr 19, 2015, 7:43 PM UTC

xamuel wrote: post #54

troels wrote: post #40

One thing I don't see discussed much in these superintelligence/singularity scenarios, is the possibility that there is an upper limit to intelligence. Maybe it simply isn't possible to be much smarter than what we already are. So maybe we build AIG, but even with all its smarts it can't improve beyond - say - Einstein level (or what have you). But surely I ca't be the first to make that suggestion, so who has discus…

Even if that's the case, quantity wins where quality fails. Imagine what you could do if you had so many clones of yourself that you could monitor every website, every forum, every person logged on IRC, 24 hours a day, giving each one of them your complete undivided attention. (Not that I'm taking an anti-AI stance. Machines do what we program them to do, that's a tautology. If there is risk, it's from user error / i…

Sure, but there is overhead of communication with a large group. One smart person can be extremely efficient, but 10 smart people are probably not 10 times as efficient. And it gets worse with scale.

Re: Stop Fearing Artificial Intelligence

#69

Post by chubot » Sun, Apr 19, 2015, 7:43 PM UTC

ikeboy wrote: post #41

chubot wrote: post #33

Earlier quoted context omitted.

Mainly because I haven't seen any evidence. Special purpose AI is not evidence. That point is debatable, but most AI practitioners agree; it's only some technologists like Elon Musk et. al. that seem to be extrapolating. Sure, humans exist, and we can already make humans by combining DNA from a man and a woman :) That method does not lead to superhuman intelligence. What makes you think we could devise some other met…

>Computers and humans are completely different things. If you watch a child grow up, this is obvious. A child can do things that are impossible for a computer, while it can't speak, read, or write. A computer can do things that no human could ever do. They are qualitatively different. Do you think humans can do things that no computer will ever do? If yes, could you give an example? If not, doesn't that imply that hu…

I think it's possible that human thought can't be simulated on a computer. It's far from obvious that it can, and this point needs justification. A lot of people are taking this for granted in their philosophical arguments.

Computers can only run Turing complete programs. Human intelligence follows the physical laws of the universe (the mechanism being unknown).

Those two substrates are different things; hence it's possible that humans can do things computers can't. Or at least the burden of proof is on the person who posits that they are equivalent.

The converse is not true: humans brains can do everything computers can, in theory and in practice. In theory we because we can compute with ours brains -- though very slowly -- and in practice because we program and control computers. We have the facility to use computers to achieve our goals, but AI algorithms don't use us to achieve their goals, to the extent that they have goals.

For a concrete task: consider producing a movie that will win an Oscar. In theory this is possible: the Godfather has a very limited number of bits, all of which can fit on a single computer many times over. The bits could just be manufactured it out of thin air.

In practice I suspect a computer would need physical agency in the world to create such a movie. It would need the ability to negotiate with Hollywood producers to raise funding; to make creative decisions to direct people who operate cameras, microphones, lighting, etc. Computers don't yet have physical agency, and I suspect they won't for quite some time, for both technological and non-technological reasons.

About simulating the universe: http://rationalwiki.org/wiki/Roko%27s_basilisk

I think this is too stupid to talk about, but it came up on Hacker News, and Yudkowsky is also referenced many times in Bostrom's "Superintelligence". It is an argument about AI that is predicated on it being possible to simulate the universe on a computer. It's almost unbelievable how this is just thrown in there and taken for granted, without justification.

I don't think strong AI is too stupid to talk about, but I believe the burden of proof is on the doomsayers, and I haven't seen arguments that don't rely on misguided and imprecise extrapolation. And I read the book "Superintelligence" to try to understand the arguments better, since people I respect like Musk and Hawking were talking about it.

As I said, practically speaking, we should be focusing on the bazillions of problems that computing and the special AI that is already here present. In the 50 year time frame, the planet could be destroyed or the population could be decimated for many other reasons.

Re: Stop Fearing Artificial Intelligence

#70

Post by icebraining » Sun, Apr 19, 2015, 7:45 PM UTC

neonbat wrote: post #61

icebraining wrote: post #42

Earlier quoted context omitted.

Up until the creation of the personal computer most machines were specialists. They were large computers designed for specific computation tasks or machines made specifically for manufacturing certain kinds of goods. Sorry, but that's just not true. Back in the 60s, way before personal computers, the were many time-sharing companies, selling general "computing resources" to different people who would dial-up into the…

Yeah I thought about that when I was writing the reply. I don't really consider that a generalized use case. Is it more generalized? Yes. Computing itself, however, can be argued to be a specialized task. My computer today doesn't do "computing" it lets me read news stories, it lets me talk on skype, it lets me send mail, it lets me have philosophical discussions on hackernews. The personal computer is so clearly a g…

PLATO had graphical terminals with touch screens and software implementing forums, message boards, online testing, e-mail, chat rooms, picture languages, instant messaging, remote screen sharing, and multiplayer games. They weren't just used by programmers by a long shot.