Leo, You are Living Under a Rock When It Comes to AI

[thenondualtankie]

AI is essentially a high dimensional digital camera. It lets you take an accurate yet finite snapshot of reality. Because all of its training data is a representative sample of (human) reality. That includes mathematics, emotion, images, sound, video, all of known science, all known logical structures and thought patterns, etc.

A camera is the same technology as an AI. But it is only able to represent reality in a very limited way: an image.

If you had a camera that could take an image of the domain of mathematics, you would have AI.

And since a camera can have arbitrarily high quality, AI can take arbitrarily better quality pictures of mathematics.

Hence why we have seen multiple examples over the last few days where AI solved major mathematical theorems:

https://www.reddit.com/r/math/comments/1uszk3d/openai_claims_to_have_proven_cycle_double_cover/

https://www.reddit.com/r/math/comments/1uxj3cy/after_openais_cdc_proof_announcement_gpt56_used_a/

https://www.reddit.com/r/math/comments/1uydg8w/gpt_56_solved_all_6_problems_from_imo_2026/

https://www.reddit.com/r/math/comments/1v1aix1/the_jacobian_conjecture_is_false_per_anthropic/

These theorems are no Riemann Hypothesis but they're significant. Just look at some of the mathematicians' disbelief in the comments.

Mathematics is just an example. AI is like a machine that can optimize for any finite, well-defined task. Chess, Go, protein folding, etc. But as it gets better, the requirement for finitude and well-definedness diminishes, and the distinction between a well-defined and ill-defined task collapses.

And that's when we achieve AGI. A finite, massive digital representation of reality that is more accurate than the sum total of mankind.

[Natasha Tori Maru]

@thenondualtankie what do you understand is Leo's take on ai?

Edited 17 hours ago by Natasha Tori Maru

[Leo Gura]

16 minutes ago, thenondualtankie said:

And that's when we achieve AGI.

No amount of data is intelligence.

[something_else]

4 minutes ago, Leo Gura said:

No amount of data is intelligence.

Yea a database isn't intelligent, but AI models aren't just databases. They are trained on data but under the hood they are abstracting and modelling that data in a way which can absolutely be considered intelligent.

The agent harness built around AI models is not just "data" either, it is a system that allows the model to think and plan.

 

[AION]

46 minutes ago, Leo Gura said:

No amount of data is intelligence.

Dude you are living with a mind and AI is mind too. I can’t believe you don’t get this. 

[thenondualtankie]

1 hour ago, Leo Gura said:

No amount of data is intelligence.

No amount of pixels is reality, but it sure as hell looks like it...

[Leo Gura]

1 hour ago, something_else said:

The agent harness built around AI models is not just "data" either, it is a system that allows the model to think and plan.

Does not matter. It is not capable of even a child's intelligence.

[Butters]

Is a PHP server intelligent? It's doing complex stuff autonomously. 

[Leo Gura]

You guys are underestimating how profound intelligence is.

[something_else]

2 minutes ago, Leo Gura said:

Does not matter. It is not capable of even a child's intelligence.

It really does matter… It can do things that no child could ever do

It’s not like intelligence is this unified thing, there are different ways to quantify and classify it

To be honest I’m not even really sure what you actually mean when you say intelligence. It feels like we’re using different definitions and that is where this disagreement stems from. What does it mean to you for something to be intelligent?

[something_else]

13 minutes ago, Butters said:

Is a PHP server intelligent? It's doing complex stuff autonomously. 

Yea, it’s just a very rigid rule-based intelligence. 

An AI model is more generally intelligent and can solve novel problems it hasn’t been explicitly instructed on how to solve before

Edited 15 hours ago by something_else

[Leo Gura]

10 minutes ago, something_else said:

What does it mean to you for something to be intelligent?

AGI.

No amount of data is AGI.

Edited 14 hours ago by Leo Gura

[thenondualtankie]

22 minutes ago, Butters said:

Is a PHP server intelligent? It's doing complex stuff autonomously. 

No. It is just a set of instructions.

But so is AI you might say? But the difference is that AI's instructions are nearly as complex as reality itself. And it gets these instructions from reality itself in the first place.

[vibv]

3 hours ago, thenondualtankie said:

AI is essentially a high dimensional digital camera. It lets you take an accurate yet finite snapshot of reality. Because all of its training data is a representative sample of (human) reality. That includes mathematics, emotion, images, sound, video, all of known science, all known logical structures and thought patterns, etc.

A camera is the same technology as an AI. But it is only able to represent reality in a very limited way: an image.

That's actually pretty accurate. But there's one caveat: You can only take an image of the past, while INTELLIGENCE is always NOW.

That's why LLMs will always be behind, they're dead intelligence—INTELLiGENCE is ALIVE.

[something_else]

18 minutes ago, Leo Gura said:

AGI.

No amount of data is AGI.

You've essentially answered "how do you define intelligence?" with "I define intelligence as intelligence." AGI is an equally vague term, so if you want to use it as your criterion for intelligence, you need to define what you mean by AGI too.

To demonstrate why that's necessary: LLM agents check most of the boxes in the standard definition of AGI below. They possess all of the traits to some degree (some in greater measure than humans, others less) so the only thing that could disqualify them is the degree to which they're better or worse than humans at each one.

[OBEler]

1 hour ago, Leo Gura said:

AGI.

No amount of data is AGI.

To some extend we already have AGI. AI can solve many jobs in economy already. its not a problem. The infrastructure is not ready yet.

[Leo Gura]

No. AGI is not a vague term. And no, we do not have anything close to AGI.

AGI is human level intelligence. No LLM is capable of that.

AGI is not able to understand things outside of its training data. LLMs cannot do that. Don't even try to act like they can.

Edited 3 hours ago by Leo Gura

[Hafiz]

LLM’s are a grift 

[Hojo]

Its not taking a snapshot of reality its taking a snapshot of maya

[something_else]

Quote

AGI is human level intelligence

Can you explain in what ways you think an LLM agent does not have human level intelligence? I don't really disagree, in some ways they are not as capable as humans. But in some ways they are more intelligent I think. Again, intelligence is not some single metric. You can be intelligent in some aspects and dumb in others.

29 minutes ago, Leo Gura said:

AGI is able to understand things outside of its training data. LLMs cannot do that. Don't even try to act like they can.

Yes, they can! The easiest proof of this is that most tech companies are using LLM coding agents on their PRIVATE codebases which the LLM has never seen during training. It loads what it thinks are the important parts of the codebase into its context and builds an internal model which it can then use to complete novel tasks within a codebase it has never seen before.

Edited 13 hours ago by something_else