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

The Case That A.I. Is Thinking

newyorker.com

Post reply on HN
Print view View original

231–240 of 1001 posts

Re: The Case That A.I. Is Thinking

#231

Post by esafak »

prmph wrote: post #214
esafak wrote: post #170

Earlier quoted context omitted.

Search is not intelligence, but synthesis is, and LLMs interpolate well. They don't invent new branches of mathematics and science yet.

You really think the reason we revere Nobel Prize winners is because they are good at interpolating?

I did not say that, but you said

> If we assembled the sum total of all published human knowledge on a storage medium and gave a computer the ability to search it extremely well in order to answer any question falling within its domain, there, you would have a Nobel Prize beating "A.I".

Which is not what LLMs do (synthesis is the core feature, not retrieval), and not how you get a Nobel Prize; many people have access to essentially all human knowledge, but have no Nobel Prize.

Re: The Case That A.I. Is Thinking

#232

Post by roadside_picnic »

goatlover wrote: post #209
roadside_picnic wrote: post #172

Earlier quoted context omitted.

> our belief that only biological entities can think Whose belief is that? As a computer scientist my perspective of all of this is as different methods of computing and we have a pretty solid foundations on computability (though, it does seem a bit frightening how many present-day devs have no background in the foundation of the Theory of Computation). There's a pretty common naive belief that somehow "thinking" is…

One argument for thinking being different from computing is that thought is fundamentally embodied, conscious and metaphorical. Computing would be an abstracted activity from thinking that we've automated with machines.

> embodied, conscious and metaphorical

Now you have 3 terms you also need to provide proper definitions of. Having studied plenty of analytical philosophy prior to computer science, I can tell you that at least the conscious option is going to trip you up. I imagine the others will as well.

On top of that, these, at least at my first guess, seem to be just labeling different models of computation (i.e. computation with these properties is "thinking") but it's not clear why it would be meaningful for a specific implementation of computation to have these properties. Are there tasks that are non-computable that are "thinkable"? And again it sounds like you're wandering into tautology land.

Re: The Case That A.I. Is Thinking

#234

Post by PaulDavisThe1st »

adamzwasserman wrote: post #219

The article misses three critical points: 1. Conflates consciousness with "thinking" - LLMs may process information effectively without being conscious, but the article treats these as the same phenomenon 2. Ignores the cerebellum cases - We have documented cases of humans leading normal lives with little to no brain beyond a cerebellum, which contradicts simplistic "brain = deep learning" equivalences 3. Most damnin…

> Conflates consciousness with "thinking"

I don't see it. Got a quote that demonstrates this?

Re: The Case That A.I. Is Thinking

#236

Post by nearbuy »

adamzwasserman wrote: post #219

The article misses three critical points: 1. Conflates consciousness with "thinking" - LLMs may process information effectively without being conscious, but the article treats these as the same phenomenon 2. Ignores the cerebellum cases - We have documented cases of humans leading normal lives with little to no brain beyond a cerebellum, which contradicts simplistic "brain = deep learning" equivalences 3. Most damnin…

Can you explain #2? What does the part of the brain that's primarily for balance and motor control tell us about deep learning?

Re: The Case That A.I. Is Thinking

#237

Post by lbrandy »

mrob wrote: post #142

I don't believe LLMs can be conscious during inference because LLM inference is just repeated evaluation of a deterministic [0] pure function. It takes a list of tokens and outputs a set of token probabilities. Any randomness is part of the sampler that selects a token based on the generated probabilities, not the LLM itself. There is no internal state that persists between tokens [1], so there can be no continuity o…

I have no idea how you can assert what is necessary/sufficient for consciousness in this way. Your comment reads like you believe you understand consciousness far more than I believe anyone actually does.

Re: The Case That A.I. Is Thinking

#238

Post by roadside_picnic »

pegasus wrote: post #180
energy123 wrote: post #154

Earlier quoted context omitted.

The point is that both are debates about definitions of words so it's extremely boring.

They can be made boring by reducing them to an arbitrary choice of definition of the word "thinking", but the question is really about weather inference is in principle as powerful as human thinking, and so would deserve to be applied the same label. Which is not at all a boring question. It's equivalent to asking weather current architectures are enough to reach AGI.

> inference is in principle as powerful as human thinking

There is currently zero evidence to suggest that human thinking violates any of the basics principles of the theory of computation nor extend the existing limits of computability.

> Which is not at all a boring question.

It is because you aren't introducing any evidence to theoretically challenge what we've already know about computation for almost 100 years now.

Re: The Case That A.I. Is Thinking

#239

Post by mrob »

lbrandy wrote: post #237
mrob wrote: post #142

I don't believe LLMs can be conscious during inference because LLM inference is just repeated evaluation of a deterministic [0] pure function. It takes a list of tokens and outputs a set of token probabilities. Any randomness is part of the sampler that selects a token based on the generated probabilities, not the LLM itself. There is no internal state that persists between tokens [1], so there can be no continuity o…

I have no idea how you can assert what is necessary/sufficient for consciousness in this way. Your comment reads like you believe you understand consciousness far more than I believe anyone actually does.

I believe consciousness needs some kind of mutable internal state because otherwise literally everything is conscious, which makes the concept useless. A rock "computes" a path to fall when you drop it but I don't believe rocks are conscious. Panpsychism is not a common belief.

Re: The Case That A.I. Is Thinking

#240

Post by czl »

crazygringo wrote: post #184
Terr_ wrote: post #132

Earlier quoted context omitted.

> I'm so baffled when I see this being blindly asserted. With the reasoning models, you can literally watch their thought process. Not true, you are falling for a very classic (prehistoric, even) human illusion known as experiencing a story : 1. There is a story-like document being extruded out of a machine humans explicitly designed for generating documents, and which humans trained on a bajillion stories humans alr…

You're claiming that the thinking is just a fictional story intended to look like it. But this is false, because the thinking exhibits cause and effect and a lot of good reasoning. If you change the inputs, the thinking continues to be pretty good with the new inputs. It's not a story, it's not fictional, it's producing genuinely reasonable conclusions around data it hasn't seen before. So how is it therefore not act…

> So how is it therefore not actual thinking?

Many consider "thinking" something only animals can do, and they are uncomfortable with the idea that animals are biological machines or that life, consciousness, and thinking are fundamentally machine processes.

When an LLM generates chain-of-thought tokens, what we might casually call “thinking,” it fills its context window with a sequence of tokens that improves its ability to answer correctly.

This “thinking” process is not rigid deduction like in a symbolic rule system; it is more like an associative walk through a high-dimensional manifold shaped by training. The walk is partly stochastic (depending on temperature, sampling strategy, and similar factors) yet remarkably robust.

Even when you manually introduce logical errors into a chain-of-thought trace, the model’s overall accuracy usually remains better than if it had produced no reasoning tokens at all. Unlike a strict forward- or backward-chaining proof system, the LLM’s reasoning relies on statistical association rather than brittle rule-following. In a way, that fuzziness is its strength because it generalizes instead of collapsing under contradiction.

Post reply on HN