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

David Deutsch On Artificial Intelligence

aeon.co

Post reply on HN
Print view View original

11–20 of 60 posts

Re: David Deutsch On Artificial Intelligence

#11

Post by jorleif »

senorgomez wrote: post #6
jorleif wrote: post #4

"only a tiny component of thinking is about prediction at all" Here I really struggle to follow his argument. It seems, he is saying that thinking is not prediction, but how does he know that? As I see it, we don't know what thinking is, so is he then saying that thinking does not feel like prediction, and therefore it cannot be prediction? It is well known that we are only aware of our very highest level of thought.…

Is there a difference between reasoning and predicting? It seems you don't think so and the author does.

I think there is a difference, but my view is that it is impossible to reason without sufficiently accurate prediction, and that the concepts we use for reasoning are largely formed based on how predictive they are. Consider playing a game where two players throw dice, and the one with the larger number wins. There is no predictability between turns, so there is no point in reasoning far into the future, unlike in chess, where the positions change slowly, one piece per turn and in predictable patterns. In chess you still don't know what the other player will do, but you can reason about it, because there is a level of predictability.

I would define reasoning as answering a question or optimization problem, e.g. What is my best move? What is a good move? This can be formalized as calculating some kind of score for a possible action, whereas prediction is just about assigning a probability value to an event.

Re: David Deutsch On Artificial Intelligence

#12

Post by ansible »

rdtsc wrote: post #9

> So, why is it still conventional wisdom that we get our theories by induction? I thought the other conventional wisdom is that we get our theories by building mental models. I throw a ball, and I can either remember how it flew in the past and use that to predict, I can also build a mental model from first principles to find out what will happen. Maybe run a short 5 second idealized simulation in my head and I can…

I thought the other conventional wisdom is that we get our theories by building mental models. I throw a ball, and I can either remember how it flew in the past and use that to predict, I can also built a mental model from first principles to find out what will happen. Maybe run a short 5 second idealized simulation in my head and I can predict what might happen. Well then I might realize that there was wind that my original model was too idealized.

My current understanding of how sports skills are developed is closer to what you first said, than the bit about 'running a simulation'.

The pro ball players have practised enough that they've seen the ball coming in at this angle, at that angle, and with this wind condition and that wind condition. And so they 'remember' where the ball is going to land, and they remember how to move to get there.

Our brains are not, as far as I know, running any kind of recognizable simulation like we would with a computer.

Re: David Deutsch On Artificial Intelligence

#13

Post by bunderbunder »

senorgomez wrote: post #6
jorleif wrote: post #4

"only a tiny component of thinking is about prediction at all" Here I really struggle to follow his argument. It seems, he is saying that thinking is not prediction, but how does he know that? As I see it, we don't know what thinking is, so is he then saying that thinking does not feel like prediction, and therefore it cannot be prediction? It is well known that we are only aware of our very highest level of thought.…

Is there a difference between reasoning and predicting? It seems you don't think so and the author does.

I think there's a strong argument to be made that the answer to this question depends on exactly what you mean by "reason" and "predict".

If you mean "predict" in the same sense as it's used in the phrase "predictive model", then the answer could be a resounding no - there's a hypothesis that the brain implements just one basic learning algorithm, and everything else is just emergent phenomena coming out of that.

Re: David Deutsch On Artificial Intelligence

#14

Post by jorleif »

enord wrote: post #5
murbard2 wrote: post #2

How on earth is Bayesianism a form of behaviorism" ? And what is it with the cult of Popper. Popperism isn't "underrated", it's considered the gold standard of epistemology despite being a vague philosophical theory full of holes. Contrast that with Solomonoff induction which has a rigorous grounding and which offers hypothesis testing and rejection as a trivial subcase.

How is Bayesianism not a form of behaviorism? The essay devotes a paragraph to explain this point allthough Deutsch assumes the reader is familiar with behaviorism and probabilistic methods used i AI.

Had he said that Bayesian decision theory with optimization of utility functions etc. was behaviorism, then I would agree, but Bayesian formulations can be used for many other things as well. At least one can propose very complex internal models, which behaviorists were criticized for not considering.

Re: David Deutsch On Artificial Intelligence

#15

Post by jorleif »

rdtsc wrote: post #9

> So, why is it still conventional wisdom that we get our theories by induction? I thought the other conventional wisdom is that we get our theories by building mental models. I throw a ball, and I can either remember how it flew in the past and use that to predict, I can also build a mental model from first principles to find out what will happen. Maybe run a short 5 second idealized simulation in my head and I can…

Why is the mental model building not seen as a special case of induction? One needs to generalize (i.e. interpolate/extrapolate) from the known instances. The mechanism may be just "replaying" known sequences, but there is still a measure of induction because one needs to decide that which simulation applies to the ball I'm seeing right now (and that what I am seeing is a ball, and that intuitive physics models of balls apply).

Re: David Deutsch On Artificial Intelligence

#16

Post by nzp »

I can just say that Deutsch is another victim of what being good at theoretical physics tends to do to one's mind. The amount of intellectual hubris and arrogance we can develop is staggering. Sheldon Cooper is not really a parody, it's what many theoretical physicists actually think (but are to socially adapted to say out loud), as in: "Penny - I'm a physicist. I have a working knowledge of the entire universe and everything it contains." Deutsch may think he knows things about "the mind" but he shows in this article that, like many who like to speculate wildly about the mind, he's basing his claim mostly on armchairing about how he feels about certain phenomena. Things he says about behaviourism are indicative that he doesn't really know what he's talking about. Or he does but is extremely bad at communicating his ideas. It would just take too long to comment on each and every empty claim this article makes. It's mostly a rehash of same old boring arguments about how special we are, this time because of our "creativity". Yawn, there is nothing special about "creativity", it's an active area of research, has been for decades, and it's actually very boring. He does make a few good points, like the one that quantum phenomena do not give rise to mental processes, but that is trivial in the sense it's obvious to anyone with some knowledge of quantum mechanics on one hand, and what neurons are and how they work on the other.

Re: David Deutsch On Artificial Intelligence

#17

Post by rdtsc »

ansible wrote: post #12
rdtsc wrote: post #9

> So, why is it still conventional wisdom that we get our theories by induction? I thought the other conventional wisdom is that we get our theories by building mental models. I throw a ball, and I can either remember how it flew in the past and use that to predict, I can also build a mental model from first principles to find out what will happen. Maybe run a short 5 second idealized simulation in my head and I can…

I thought the other conventional wisdom is that we get our theories by building mental models. I throw a ball, and I can either remember how it flew in the past and use that to predict, I can also built a mental model from first principles to find out what will happen. Maybe run a short 5 second idealized simulation in my head and I can predict what might happen. Well then I might realize that there was wind that my…

Interesting. I do visualize in my head how kinetic energy depletes as it goes higher and potential energy increases. It is like a bubble that shrinks. I see motion vectors and how they change.

Other examples are a conflict between 3 people, kind of see a graph of relationships with different colored arrows and how they change.

Re: David Deutsch On Artificial Intelligence

#18

Post by rdtsc »

jorleif wrote: post #15
rdtsc wrote: post #9

> So, why is it still conventional wisdom that we get our theories by induction? I thought the other conventional wisdom is that we get our theories by building mental models. I throw a ball, and I can either remember how it flew in the past and use that to predict, I can also build a mental model from first principles to find out what will happen. Maybe run a short 5 second idealized simulation in my head and I can…

Why is the mental model building not seen as a special case of induction? One needs to generalize (i.e. interpolate/extrapolate) from the known instances. The mechanism may be just "replaying" known sequences, but there is still a measure of induction because one needs to decide that which simulation applies to the ball I'm seeing right now (and that what I am seeing is a ball, and that intuitive physics models of ba…

> Why is the mental model building not seen as a special case of induction? One needs to generalize (i.e. interpolate/extrapolate) from the known instances.

Maybe you are right. Now, the problem is that my explanation pushes the problem down because now I am saying, "I don't know how this model simulator works. I throw a bunch of objects or concepts in a box and they evolve according to some rules, I can see where thing will end up in the future". But I don't have a fine grain understanding of what the simulator is built, or if it has frames or steps that execute and how it combines the rules underneath. That is perhaps in the unconscious. So I haven't solved the problem but just exposed the highest level of reasoning that I am aware of. There is some assembly running that I am not aware.

Re: David Deutsch On Artificial Intelligence

#19

Post by murbard2 »

jorleif wrote: post #14
enord wrote: post #5

Earlier quoted context omitted.

How is Bayesianism not a form of behaviorism? The essay devotes a paragraph to explain this point allthough Deutsch assumes the reader is familiar with behaviorism and probabilistic methods used i AI.

Had he said that Bayesian decision theory with optimization of utility functions etc. was behaviorism, then I would agree, but Bayesian formulations can be used for many other things as well. At least one can propose very complex internal models, which behaviorists were criticized for not considering.

Agreed. Besides, behaviorism was abandoned in psychology because it was crazy, not because there's something wrong with reinforcement learning.

It was treating the human mind as a black box trying to pretend we have no insight in human thoughts. It was trying to avoid antropomorphizing people as Morgenbesser put it. In fact, quite the opposite, I'd attribute this blatant idiocy to Popperism. After all, introspection isn't testable or falsifiable in a popperian sense. There's no room for the concept of models in Popperianism. There are just big black box ideas that you test and are falsified or not.

Re: David Deutsch On Artificial Intelligence

#20

Post by ArbitraryLimits »

nzp wrote: post #16

I can just say that Deutsch is another victim of what being good at theoretical physics tends to do to one's mind. The amount of intellectual hubris and arrogance we can develop is staggering. Sheldon Cooper is not really a parody, it's what many theoretical physicists actually think (but are to socially adapted to say out loud), as in: "Penny - I'm a physicist. I have a working knowledge of the entire universe and e…

Speaking as someone with a physics degree, I like to quote Peter Thiel (for once):

> This is why physics Ph.D’s are notoriously difficult to work with; because they know fundamental things, they think they know all things.

- http://blakemasters.com/post/22866240816/peter-thiels-cs183-...

Post reply on HN