Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Physicist proposes new way to think about intelligence

insidescience.org

Post reply on HN
Print view View original

81–90 of 145 posts

Re: Physicist proposes new way to think about intelligence

#81

Post by wamatt »

cscurmudgeon wrote: post #34
Steuard wrote: post #27

I haven't read the original paper, but (speaking as a physicist) I'm always skeptical when physicists leap into extremely different disciplines and claim to have a grand insight that its specialists have never considered before. It could always be true: physics really does have a lot to contribute in many areas! But the burden of proof is very, very much on the physicists to justify the relevance of their approach. I…

Welcome to modern physics, where physicists do unfalsifiable and imprecise things based on bad philosophy. Anything goes as long as it gets you attention anywhere. https://news.ycombinator.com/item?id=5562156 HNers need to more critical of Science articles. This is where we need good criticism not on some poor hacker's pet project. Edit 1 "Additionally, a company he founded is exploring commercial applications of the…

It's turtles all the way down, until you reach Philosophy

Re: Physicist proposes new way to think about intelligence

#82

Post by pjscott »

wamatt wrote: post #81
cscurmudgeon wrote: post #34

Earlier quoted context omitted.

Welcome to modern physics, where physicists do unfalsifiable and imprecise things based on bad philosophy. Anything goes as long as it gets you attention anywhere. https://news.ycombinator.com/item?id=5562156 HNers need to more critical of Science articles. This is where we need good criticism not on some poor hacker's pet project. Edit 1 "Additionally, a company he founded is exploring commercial applications of the…

It's turtles all the way down, until you reach Philosophy

... At which point it's still turtles, but nobody can really agree on what a turtle is, or what it means for something to be turtles, or what it means for something to be anything, including (but not limited to) turtles, or what it means for something to mean something. By the time the subject is settled -- if it ever is settled -- the turtle is dead. And so is God.

(For some definitions of "turtle", "dead", "God", and most especially "is".)

Re: Physicist proposes new way to think about intelligence

#83

Post by mikhailfranco »

As referenced in the paper, Verlinde proposed that gravity is an entropic force:

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

I highly recommend Verlinde's brilliant paper, it's an easy read:

http://arxiv.org/abs/1001.0785

The theory remains controversial, but it's a valuable attempt to put holographic principles center stage.

Re: Physicist proposes new way to think about intelligence

#84

Post by kespindler »

thibauts wrote: post #58
dsowers wrote: post #33

Earlier quoted context omitted.

Yeah, I had to parse the article a couple times because the wording is wrong. It seems to me that entropica is trying to minimize the entropy of the system. Life is all about low entropy.

Indeed. I'd say intelligence tries to keep its environment in a state it can predict.

I think the trick to understanding this is that, in order for an intelligent being to maintain it's internal homeostasis, it must (in an entropic sense) really screw up it's surrounding environment. Thus, if you look at the whole system, the intelligent being is maximizing entropy. Phrased another way: Because it wants to maximize it's own homeostasis, it has to work extra hard to achieve this. Work creates entropy when looking outside the whole system, and bam - you have entropy maximization.

I'm not sure I'm right, but this logic works for me...

Re: Physicist proposes new way to think about intelligence

#85

Post by Matti »

pitchups wrote: post #35
conroe64 wrote: post #5

http://www.entropica.com/ has a very nice, short, video demonstrating the AI.

Completely blown away after watching that video by the sheer range of tasks this AI can accomplish on its own. Learning to walk, balance a stick, using tools, co-operating, playing pong, and...even buying stocks low and selling high! And all of this without being programmed or given a goal to do so. If true this could be a huge breakthrough indeed. Would really like to learn more about the core technology and will be…

Where do you see it learning to walk? You can't equate the inverted pendulum problem with upright walking. It is neat if it can choose it's own reference level ("goal") for a classical control problem -- but that's a long way from "learning to walk".

Re: Physicist proposes new way to think about intelligence

#87

Post by scotty79 »

thibauts wrote: post #58
dsowers wrote: post #33

Earlier quoted context omitted.

Yeah, I had to parse the article a couple times because the wording is wrong. It seems to me that entropica is trying to minimize the entropy of the system. Life is all about low entropy.

Indeed. I'd say intelligence tries to keep its environment in a state it can predict.

Research just observes that if you try to keep your environment in the state in which you have maximum number of possible futures you can choose from by acting in specific way, then your behavior leading to this state seems intelligent.

Re: Physicist proposes new way to think about intelligence

#88

Post by scotty79 »

klipt wrote: post #80
svachalek wrote: post #50

Earlier quoted context omitted.

It's not "predicting" the future, it's "maximizing" the number of potential futures. In chess, the more moves you have available the better you are doing, generally. But another measure (probably better although I'm not a chess expert) is how few moves your opponent has available. Strictly speaking it sounds like this strategy would emphasize keeping future moves available for both you and your opponent. This is a mo…

Yeah, seems to me their idea can be summarized as "in the absence of a specific goal, keep options open". Which is generally good advice, but only "in the absence of a specific goal".

I think that's what is different between machine and life. Machine always has specific goal. Life never has.

Re: Physicist proposes new way to think about intelligence

#89

Post by klipt »

scotty79 wrote: post #88
klipt wrote: post #80

Earlier quoted context omitted.

Yeah, seems to me their idea can be summarized as "in the absence of a specific goal, keep options open". Which is generally good advice, but only "in the absence of a specific goal".

I think that's what is different between machine and life. Machine always has specific goal. Life never has.

Well, most life has a tendency to reproduce (inherited from ancestors who all reproduced). Aka evolution.

Re: Physicist proposes new way to think about intelligence

#90

Post by scotty79 »

colanderman wrote: post #12

This isn't just about keeping options open. Entropy can be seen as a measure of available useful energy. It seems that it would be natural for human intelligence to have evolved to maintain as much useful energy as possible, since that would no doubt aid survival. If that's true, I don't think it's as much that the developed process mimic intelligence, as much as that both (human) intelligence and this process act to…

I think this research is the missing part of the puzzle of constructing truly intelligent (even sentient) artificial life.

We can't do it the way evolution did it because evolution does not work on individuals and we don't have enough computing power to simulate whole population of artificial human sized brains and let them figure out on their own that they should maximize their available useful resources while still reproducing and evolving towards better architecture. That would lead to intelligence same way real evolution reached it but we neither have power nor time for that. Besides, even if we succeeded ... do we really want to share the world with thousands AI-s all trying to maximize useful resources available to them? Competition breeds innovation but also breeds war.

If you design AI and give it a specific goal you never get intelligence, you are just getting useful machine that achieves the goal brilliantly. With this discovery we finally know how to set the goal to be "act intelligent"

If you artificially set the goal to be to maximize the possible futures that the AI can make real from the present point we can get entity that behaves in a way we recognize as intelligent.

We just have to remember to be really sure that this AI models world correctly enough to know that more potential futures could be realized by keeping animals (including humans) around and ... you now ... not make them suffer.

Post reply on HN