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

AGI is Mathematically Impossible 2: When Entropy Returns

philarchive.org

Post reply on HN
Print view View original

241–250 of 437 posts

Re: AGI is Mathematically Impossible 2: When Entropy Returns

#241

Post by cess11 »

Merleu-Ponty would be a less wasteful path to this kind of conclusion, who was more or less introduced to the US by Hubert Dreyfus, infamously contrarian while at MIT during an earlier phase in AI fashion and author of books such as What Computers Can't Do and What Computers Still Can't Do.

It's a trivial observation that binary CPU:s and memory systems are fundamentally different from ugly, analog, bags of mostly water. To force binary systems to perform a human-like mimicry necessarily entails a lot of emulation, and to emulate not just a strictly limited portion of a human would use a lot more resources than a human would.

Re: AGI is Mathematically Impossible 2: When Entropy Returns

#242

Post by sgt101 »

vidarh wrote: post #229
bloqs wrote: post #203

Earlier quoted context omitted.

could you explain for a layman

I'm not sure if this will help, but happy to elaborate further: The set of Turing computable functions is computationally equivalent to the lambda calculus, is computationally equivalent to the generally recursive functions. You don't need to understand those terms, only to know that these functions define the set of functions we believe to include all computable functions . (There are functions that we know to not b…

Compute functions != Intelligence though.

For example learning from experience (which LLMs cannot do because they cannot experience anything and they cannot learn) is clearly an attribute of an intelligent machine.

LLMs can tell you about the taste of a beer, but we know that they have never tasted a beer. Flight simulators can't take you to Australia, no matter how well they simulate the experience.

Re: AGI is Mathematically Impossible 2: When Entropy Returns

#244

Post by cess11 »

sgt101 wrote: post #237
bubblyworld wrote: post #235

I find the mathematics in this paper a little incoherent so it's hard to criticise it on those grounds - but on a charitable read, something that sticks out to me is the assumption that AGI is some fixed total computable function from the fixed decision domain to a policy. AIs these days autonomously seek information themselves. Much like living things, they are recycling entropy and information to/from their environ…

Ok - where do AIs put the information that they "seek" from the internet?

I suspect there's a harsher argument to be made regarding "autonomous". Pull the power cord and see if it does what a mammal would do, or if it rather resembles a chaotic water wheel.

Re: AGI is Mathematically Impossible 2: When Entropy Returns

#245

Post by davedx »

sgt101 wrote: post #237
bubblyworld wrote: post #235

I find the mathematics in this paper a little incoherent so it's hard to criticise it on those grounds - but on a charitable read, something that sticks out to me is the assumption that AGI is some fixed total computable function from the fixed decision domain to a policy. AIs these days autonomously seek information themselves. Much like living things, they are recycling entropy and information to/from their environ…

Ok - where do AIs put the information that they "seek" from the internet?

Into their short term memory (context). Some information is also stored in long term memory (user store)

Re: AGI is Mathematically Impossible 2: When Entropy Returns

#246

Post by throw310822 »

TheOtherHobbes wrote: post #234
vidarh wrote: post #229

Earlier quoted context omitted.

I'm not sure if this will help, but happy to elaborate further: The set of Turing computable functions is computationally equivalent to the lambda calculus, is computationally equivalent to the generally recursive functions. You don't need to understand those terms, only to know that these functions define the set of functions we believe to include all computable functions . (There are functions that we know to not b…

What program would a Turing machine run to spontaneously prove the incompleteness theorem? Can you prove such a program may exist?

An accurate-enough physical simulation of Kurt Gödel's brain.

Such a program may exist- unless you think such a simulation of a physical system is uncomputable, or that there is some non-physical process going on in that brain.

Re: AGI is Mathematically Impossible 2: When Entropy Returns

#247

Post by bubblyworld »

sgt101 wrote: post #237
bubblyworld wrote: post #235

I find the mathematics in this paper a little incoherent so it's hard to criticise it on those grounds - but on a charitable read, something that sticks out to me is the assumption that AGI is some fixed total computable function from the fixed decision domain to a policy. AIs these days autonomously seek information themselves. Much like living things, they are recycling entropy and information to/from their environ…

Ok - where do AIs put the information that they "seek" from the internet?

I can see what you are getting at but consider:

I had an experience the other day where claude code wrote a script that shelled out to other LLM providers to obtain some information (unprompted by me). More often it requests information from me directly. My point is that the environment itself for these things is becoming at least as computationally complex or irreducible (as the OP would say) as the model's algorithm, so there's no point trying to analyse these things in isolation.

Re: AGI is Mathematically Impossible 2: When Entropy Returns

#248

Post by danieltanfh95 »

This is consistent with AI usage patterns that people now internalise: start a new context everytime you have a new task. LLMs suck at dealing with context poisoning, intended or not, and the more information they have access to or involved in the conversation, the worse AI performs for its cognitive function.

Re: AGI is Mathematically Impossible 2: When Entropy Returns

#249

Post by IanCal »

This is atrocious.

> There exists a class of questions in life that appear remarkably simple in structure and yet contain infinite complexity in their resolution space. Consider the familiar or even archetypal inquiry: "Darling, please be honest: have I gained weight?" Now, let’s observe what happens when an AI system - equipped with state-of-the-art natural language processing, sentiment analysis, and social reasoning - attempts to navigate this question

Yes, let's.

None of the systems go into an infinite loop. We simply don't let them.

Here's o3 https://chatgpt.com/share/68591a21-de4c-8002-94cd-bf6cc5b269...

That's handled with dramatically better tact than the author

> (Note to my wife, should she read this: This is a purely theoretical example for an algorithmically unsolvable riddle, love. You look wonderful, as you always did. And to the reader: No, I am not trying to find a way out of the problem I just got myself into here: I am neither stupid nor suicidal. So, you can conclude that my wife indeed is truly beautiful, for I wouldn't be so dumb to pick that example if she wasn't. And yes, I know: You now ask yourself if this sentence WAS my way out... tricky, no?)

It is the height of laziness or arrogance to write about how AI "can't do X" without simply trying. The models, particularly things like o3 with searching are extremely good at lots of things.

Post reply on HN