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

131–140 of 145 posts

Re: Physicist proposes new way to think about intelligence

#131

Post by sramsay »

calhoun137 wrote: post #104
BenoitEssiambre wrote: post #96

Earlier quoted context omitted.

The article is kind of written in gibberish. It also implies that the idea of linking thermodynamics to intelligence is novel when a lot of experts actually see it as a fundamental of thermodynamics and the universe. It's even possible that this physicist is naive about the whole debate an thinks that he discovered this. However while the idea is not new, it does seem like he made a software implementation of some so…

Just so you know, before I started spending all of my time programming I was a physicist, and statistical physics was my best subject. Anyway, I kind of object to your definition of intelligence as "the ability to predict the unknown". Of course it comes down to personal preference, but I feel that when it comes to making AI, a better definition is something along the lines of "intelligence is the ability to make goo…

"I feel that when it comes to making AI, a better definition is something along the lines of "intelligence is the ability to make good choices when faced with a decision"

I love it. Mine is even more constructivist (and Wittgensteinian). Something like "intelligence is the ability to detect patterns and act accordingly."

The problem, as many philosophers have noted, is that such definitions -- in fact, most definitions -- make it hard to argue that the thermostat is not intelligent (or conscious, or sentient). And that, of course, is at the heart of Turing's most famous argument.

I, myself, find it hard to argue my way around what is essentially a "walks like a duck, talks like a duck" argument when it comes to these matters -- which is to say, defining something in terms of its activities, behaviors, likenesses. That perhaps tilts practical AI toward the statistical, though, and it's hard for me to believe that that's the whole story.

I have never presented the arguments back-and-forth on this issue without watching my students collapse into confusion and desperation.

Re: Physicist proposes new way to think about intelligence

#132

Post by snowwrestler »

Houshalter wrote: post #59
snowwrestler wrote: post #31

Earlier quoted context omitted.

Reproducing captures more future states than not reproducing, since it allows some of your collected information to extend past your death.

You can explain almost any process with this theory, and that's why it's useless. Trying to "capture more future states" is not the reason certain traits evolve, they evolve simply because they are more likely to survive and reproduce.

Why any living thing reproduces at all is a much more fundamental question than e.g. why men have nipples or snakes don't have legs.

Re: Physicist proposes new way to think about intelligence

#133

Post by Houshalter »

snowwrestler wrote: post #132
Houshalter wrote: post #59

Earlier quoted context omitted.

You can explain almost any process with this theory, and that's why it's useless. Trying to "capture more future states" is not the reason certain traits evolve, they evolve simply because they are more likely to survive and reproduce.

Why any living thing reproduces at all is a much more fundamental question than e.g. why men have nipples or snakes don't have legs.

Because things that do reproduce quickly out-reproduce and out-compete things that don't.

Re: Physicist proposes new way to think about intelligence

#134

Post by Evbn »

sramsay wrote: post #131
calhoun137 wrote: post #104

Earlier quoted context omitted.

Just so you know, before I started spending all of my time programming I was a physicist, and statistical physics was my best subject. Anyway, I kind of object to your definition of intelligence as "the ability to predict the unknown". Of course it comes down to personal preference, but I feel that when it comes to making AI, a better definition is something along the lines of "intelligence is the ability to make goo…

"I feel that when it comes to making AI, a better definition is something along the lines of "intelligence is the ability to make good choices when faced with a decision" I love it. Mine is even more constructivist (and Wittgensteinian). Something like "intelligence is the ability to detect patterns and act accordingly." The problem, as many philosophers have noted, is that such definitions -- in fact, most definitio…

The great personal insight is when one finally discards prejudice and bigotry and realizes that a thermostat is intelligent, and that a neuron is a "molecule" of intelligence.

Re: Physicist proposes new way to think about intelligence

#135

Post by Evbn »

calhoun137 wrote: post #113
ac500 wrote: post #110

Earlier quoted context omitted.

> "I feel that when it comes to making AI, a better definition is something along the lines of "intelligence is the ability to make good choices when faced with a decision"." What defines a good choice? What defines a bad choice? Saying "A true machine intelligence is one that makes good decisions" is a tautology -- it doesn't define anything. Rather, it just reframes one vague question into another. Intelligence is…

For a rational agent, a decision is good or bad based on it's preferences, which are described mathematically as part of the definition of the rational agent. So I don't believe it's tautological. I do agree with your statement that intelligence is difficult to describe. And for the record I never claimed the rational agent approach will achieve "true machine intelligence". Frankly, I don't even know what "true" mach…

Thank you for knocking down the Gödel canard. People who say humans are not similar able have a vague faith that humans are far smarter than they really are.

The existence of "Uhprovably true statements" is less of a weakness in logic's ability to organize the universe, and more that there are always less meaningful yet constructible statements, like "This statement is not provable."

Re: Physicist proposes new way to think about intelligence

#136

Post by ac500 »

Evbn wrote: post #135
calhoun137 wrote: post #113

Earlier quoted context omitted.

For a rational agent, a decision is good or bad based on it's preferences, which are described mathematically as part of the definition of the rational agent. So I don't believe it's tautological. I do agree with your statement that intelligence is difficult to describe. And for the record I never claimed the rational agent approach will achieve "true machine intelligence". Frankly, I don't even know what "true" mach…

Thank you for knocking down the Gödel canard. People who say humans are not similar able have a vague faith that humans are far smarter than they really are. The existence of "Uhprovably true statements" is less of a weakness in logic's ability to organize the universe, and more that there are always less meaningful yet constructible statements, like "This statement is not provable."

> "People who say humans are not similar able have a vague faith that humans are far smarter than they really are."

Prenote: I apologize in advance if my entire response is off-topic, because my English parser was completely unable to take anything meaningful from this sentence.

When you have a system of axioms, both consistent and complete, which proves all useful properties of arithmetic and algebra, let us know.

Whether or not you are able to understand the deepness of connection between Godel's theorems, formal systems, self-similarity, and intelligence, you can't deny that the more statistical / 'organic' AIs are overwhelmingly more successful than rigidly logical ones. For example, try implementing voice recognition analytically, and let us know your results (have fun with that).

The point is that although some people would like to believe otherwise, human brains (and nonhuman brains) are NOT logical reasoning machines. Humans are not rational agents. It still surprises me how many people are desperately grasping to rational agents and similar frameworks to try to build general intelligences entirely from logic. I point out Godel's theorems as one type of insight into why this road will lead you nowhere. You're free to pursue it as much as you like, just as animals are free to run in circles chasing their tail.

You can't build an AI capable of thinking outside the box, by specifically building that AI in the box of a particular logical framework. Logic itself has limitations. Logic is powerful, but not entirely in of itself -- logic is powerful as a tool extended by the mind of its creator(s). The more meaningful question in AI, is how do we create logical concepts to begin with?

Re: Physicist proposes new way to think about intelligence

#137

Post by tinco »

dusklight wrote: post #112
tinco wrote: post #92

Earlier quoted context omitted.

The idea is that these models could be extended to really big complex problems with the same basic principle. Therefore, it seems that intelligence of arbitrary complexity can be achieved with these algorithms. I don't think that claim is very bold at all, it is merely an observation. The idea that the algorithm tries to maintain entropy is very much like life as we know it. Life as we know it are mechanisms that des…

It's pretty clear that we are not the only animals with brains, so there are other animals who use their brains to make decisions, such as how to hunt and when to sleep and so on. We make what is in our opinion more sophisticated decisions, but that arguably is a human-centric conceit. There are ant colonies that span continents, and they aren't building any nuclear weapons. Are we really sure we are more intelligent…

The brain is an instrument that allows a species to be more adaptable to new environments, of course there are other ways to achieve adaptability, being simple and generic is one of them. Being a symbiote/parasite of an adaptable species (like bacteria are of us) is another.

I think the battle between complex organisms and simple organisms is an ongoing one. At the moment we rely on the simple organisms on our insides to protect us against the simple organisms on the outside. If there is any threat to the survival of our species it lies there. A nuclear holocaust or a few degrees of global warming isn't going to do it, unless you define 'our species' as 'inhabitants of new york'.

Wether we are more intelligent that some other intelligent animals is dependent on how you measure intelligence. If you define it as having many offspring in many continents, I'm not sure ants would win, but I'm not sure humans would either.

Re: Physicist proposes new way to think about intelligence

#138

Post by leot »

cscurmudgeon wrote: post #125
leot wrote: post #124

I came here hoping to find intelligent commentary on the article. Instead, it's mostly uninformed criticism. The most promoted comments are basically "I didn't carefully read or try hard to understand this paper, but everything I think I know up to now says that it can't be very interesting." I don't really understand the paper either. But given that it was reviewed and published in a good journal, and given that it'…

Science requires constant criticism especially when people claim grand things. Don't assume people here are idiots and have not read the paper. Every year there are dozens of papers claiming a grand theory of AI.

I'm not assuming that. In fact, I assumed the opposite and it was after I read most of the comments that it became evident that most people here weren't engaging with the article in any kind of substantive or even remotely sympathetic way (and, believe it or not, good academic reading requires that one adopt the most charitable interpretation of the author's claims -- at least to begin with).

And, please don't assume that I don't understand how science or AI theories work, either. If you email me I'd be happy to explain my bonafides.

Re: Physicist proposes new way to think about intelligence

#139

Post by rthomas6 »

solistice wrote: post #129
phillmv wrote: post #52

Earlier quoted context omitted.

Whelp, that's completely unreadable to me. Undergraduate computer science fail. I got as far as "the model of this physical process of entropy can act as a search function for lower entropic states, which given some poking, can resemble problem solution states". It's the poking part that is problematic. >To better understand this classical-thermal form (11) of causal entropic forcingwe simulated its effect [12,13] on…

phill, this is hacker news, and if someone finds you're wildly unqualified to comment on something, don't worry, he'll be vocal about it and explain exactly why he thinks you're unqualified to comment about it. That's why I love this site. On any similar site aggregator, I'd get cat pictures and people telling me my opinions are stupid because my face is stupid. More towards the topic, I'd say these kind of articles…

Then there's lesswrong.com, where I've never felt qualified to even comment at all.

Re: Physicist proposes new way to think about intelligence

#140

Post by SittingDuck »

technotony wrote: post #7

This is fascinating. It proposes that intelligence is about maximizing the control we have over future events, ie maximize the entropy of the system. Intuitively this aligns well with how I think about many strategic options or inflections in life/business - you try to take the path which maximizes future options. Lots of interesting applications, will be interested to see how well they manage to communicate this wor…

Your analysis is not fully accurate, if I understand the paper correctly; I believe you do not maximize the actual8 entropy of the system. You maximize the potential* future entropy of the system. That's something entirely different.

There is actually a line of research where channel capacity of the actuation/perception channel is being used to model that (it's also named "empowerment", in analogy to the social science term, but it is measured in terms of information theoretical quantities). Here, one does not only model the entropy of the future, but in fact that amount of entropy about the future that the agent can systematically control. Or, more precisely, how well the agent is able to systematically control its future (which allows it maximal future entropy - but only entropy it can actually itself generate!)

See e.g. Klyubin et al. 2005, 2008, Capdepuy et al. 2007 and later, Jung et al. 2011, Salge et al. 2012. Indeed, maximizing the control over potential futures (empowerment) seems to work well for a range of scenarios, including the discovery of points of high centrality in mazes and graphs, as well as pole-balancing, acrobot balancing, bicycle balancing examples, in short, survival scenarios; also for discovering new object manipulation modes, driving sensor evolution models, self-organizing multiagent collective behaviour and a few other cases. It does not work well for systems with prespecified goals, systems with funnels/bottleneck transitions, in short, for cases where you have to relinquish potential futures to commit to a decision. This needs to be treated with a different principle.

The information-theoretic treatment takes into account action routes with controllable dynamics, it will avoid state space regions with more noise and thus less controllability. Importantly: it is the "potential" to do something, it does not enforce to actually carry out the option. Having the option is all that counts. Like in chess, the "threat" is more effective than the "carrying out".

Disclaimer: I am one of the authors of the empowerment work.

Post reply on HN