Surprisingly fast AI-generated kernels we didn't mean to publish yet

And other companies have existed for hundreds of years and had thousands of people work for them and never even made $100M.

* replying with “I don’t know” a lot more often

* consistent responses based on the accessible corpus

* far fewer errors (hallucinations)

* being able to beat Pokémon reliably and in a decent time frame without any assistance or prior knowledge about the game or gaming in general (Gemini 2.5 Pro had too much help)

Your question can be rephrased to “what would an actual difference look like.”

However, what you are asking underneath that, is a mix of “what is the difference” and “what is the PRACTICAL difference in terms of output”

Or in other words, if the output looks like what someone with understanding would say, how is it meaningfully different.

—-

Humans have a complex model of the world underlying their thinking. When I am explaining this to you, you are (hopefully) not just learning how to imitate my words. You are figuring out how to actually build a model of an LLM, that creates intuitions / predictions of its behavior.

In analogy terms, learning from this conversation, (understanding) is to create a bunch of LEGO blocks in your head, which you can then reuse and rebuild according to the rules of LEGO.

One of the intuitions is that humans can hallucinate, because they can have a version of reality in their head which they know is accurate and predicts physical reality, but they can be sick/ill and end up translating their sensory input as indicating a reality that doesn’t exist. OR they can lie.

Hallucinations are a good transition point to move back to LLMs, because LLMs cannot actually hallucinate, or lie. They are always “perceiving” their mathematical reality, and always faithfully producing outputs.

If we are to anthropomorphize it back to our starting point about “LLMs understand”, this means that even when LLMs “hallucinate” or “lie”, they are actually being faithful and honest, because they are not representing an alternate reality. They are actually precisely returning the values based on the previous values input into the system.

“LLMs understand” is misleading, and trojans in a concept of truth (therefore untruth) and other intuitions that are invalid.

—-

However, understanding this does not necessarily change how you use the LLMs 90% of the time, it just changes how you model them in your head, resulting in a higher match between observer reality and your predictive reality.

For an LLM this makes not difference, because its forecasting the next words the same way.

No. You have an initial prompt that is vague, and then you have another prompt that is more specific.

- "draw me an automobile"

- "here's a picture of an ambulance."

- "could you make it a convertible instead? Perhaps green."

- "ok, here's a picture of a jaguar e-type".

My favorite thing about LLMs is that they can convincingly tell me why I'm wrong or how I could think about things differently, not for ideas on the order of sentences and paragraphs, but on the order of pages.

My second favorite thing is that it is amazingly good at deconstructing manipulative language and power tactics. It is scary good at developing manipulation strategies and inferring believable processes to achieve complex goals.

And if that is so, didn't you also "just" express an opinion? Would your own contribution to the discussion pass your own test?

You might have overlooked that I provided extensive arguments all around in this thread. Please reconsider.

Claiming anything else requires a proof.

An LLM can pass many tests, it is indistinguishable from someone who understands the subject.

Indistinguishable does not imply that the processes followed match what a human is doing when it understands a subject.

I use this when I think of humans learning - humans learn the most when they are playing. They try new things, explore ideas and build a mental model of what they are playing with.

To understand something, is to have a mental model of that thing in ones head.

LLMs have models of symbol frequency, and with their compute, are able to pass most tests, simply because they are able to produce chains of symbols that build on each other.

However, similar to rote learning, they are able to pass tests. Not understand. The war is over the utilitarian point “LLMs are capable of passing most tests”, and the factual point “LLMs dont actually understand anything”.

This articulation of the utilitarian point is better than the lazier version which says “LLMs understand”, and this ends up anthropomorphizing a tool, and creating incorrect intuitions of how LLMs work, amongst other citizens and users.

Permit my likely inaccurate illustration: You’re pretty sure 2 + 2 is 4, but there are several questions you could ask: are any of the numbers negative, are they decimals, were any numbers left out? Most of those questions are things you’ve learned to ask automatically, without thinking about it, because you know they’re important. But because the answer matters, you check your work by writing out the equation. Then, maybe you verify it with more math; 4 ÷ 2 = 2. Now you’re more confident the answer is right.

An LLM doesn’t understand math per se. If you type “2 + 2 =”, the model isn’t doing math… it’s predicting that “4” is the next most likely token based on patterns in its training data.

“Thinking” in an LLM is like the model shifting mode and it starts generating a list of question-and-answer pairs. These are again the next most likely tokens based on the whole context so far. “Reasoning” is above that: a controlling pattern that steers those question-and-answer sequences, injecting logic to help guide the model toward a hopefully more correct next token.