The LLMentalist Effect

I've read this already, also from a HN link. It's a good piece to explain why people believe that large language models reached some sort of intelligence, instead of being glorified text generators. (Sometimes a glorified text generator is damn useful, but we shouldn't pretend that it's something else.)

I'll list here a few fallacious arguments that you see being used to defend that LLMs are intelligent, as well as some rather grumpy answers.

Now let me see the HN comments. Given the subject it's a shitfest, but I'm low-key in the mood for popcorn.

This is from 2023 and is clearly dated. It is mildly interesting to notice how quickly things changed since then. Nowadays models can solve original math puzzles much of the time and it is harder to argue they cannot reason when we have access to R1, o1, and o3-mini.

It is not really dated. People obsess over the metric (in this case, maths puzzles) without considering what the metric is supposed to gauge (ability to reason). Even DeepSeek and GPT o4-mini are still vomiting brainfarts here and there, and both are from my experience the best costless models there. ("Costless" for the user. Not for the environment. I just don't want to use the word "free" as it's ambiguous here.)

I was hoping it was talking about how it can resonate with users using those techniques. Or some experiments to prove the point. But it is not even that. // There is nothing of substance in this and it feels like the author has a grudge against LLMs.

Refer to "You must have something against LLMs."

Plus typical "Redditors LARP as 5UP4 1337 H4x0rz" behaviour - "waah, it's missing [thing]!"

Turns out that identifying problems and solving them are two different skills, and people who are good with one aren't necessarily good with the other. So it's completely fine to point out a problem (e.g. suckers believing LLMs are smart) without necessarily providing a solution.

This feels rather forced. The article seems to claim both that LLMs don't actually work, it is all an illusion and that of course the LLMs know everything, they stole all our work from the last 20 years by scraping the internet and underpaying people to produce content. If it was a con, it wouldn't have to do that. Or in other words, if you had a psychic who actually memorized all biographies of all people ever, they wouldn't need their cons

What doesn't actually work is the reading comprehension of this user.

You should try the arc agi puzzles yourself, and then tell me you think these things aren't intelligent / https://arcprize.org/blog/openai-o1-results-arc-prize / I wouldn't say it's full agi or anything yet, but these things can definitely think in a very broad sense of the word

Goodhart's Law.

1 comments

I've read this already, also from a HN link. It's a good piece to explain why people believe that large language models reached some sort of intelligence, instead of being glorified text generators. (Sometimes a glorified text generator is damn useful, but we shouldn't pretend that it's something else.)
I'll list here a few fallacious arguments that you see being used to defend that LLMs are intelligent, as well as some rather grumpy answers.

Now let me see the HN comments. Given the subject it's a shitfest, but I'm low-key in the mood for popcorn.
This is from 2023 and is clearly dated. It is mildly interesting to notice how quickly things changed since then. Nowadays models can solve original math puzzles much of the time and it is harder to argue they cannot reason when we have access to R1, o1, and o3-mini.
It is not really dated. People obsess over the metric (in this case, maths puzzles) without considering what the metric is supposed to gauge (ability to reason). Even DeepSeek and GPT o4-mini are still vomiting brainfarts here and there, and both are from my experience the best costless models there. ("Costless" for the user. Not for the environment. I just don't want to use the word "free" as it's ambiguous here.)
I was hoping it was talking about how it can resonate with users using those techniques. Or some experiments to prove the point. But it is not even that. // There is nothing of substance in this and it feels like the author has a grudge against LLMs.
Refer to "You must have something against LLMs."
Plus typical "Redditors LARP as 5UP4 1337 H4x0rz" behaviour - "waah, it's missing [thing]!"
Turns out that identifying problems and solving them are two different skills, and people who are good with one aren't necessarily good with the other. So it's completely fine to point out a problem (e.g. suckers believing LLMs are smart) without necessarily providing a solution.
This feels rather forced. The article seems to claim both that LLMs don't actually work, it is all an illusion and that of course the LLMs know everything, they stole all our work from the last 20 years by scraping the internet and underpaying people to produce content. If it was a con, it wouldn't have to do that. Or in other words, if you had a psychic who actually memorized all biographies of all people ever, they wouldn't need their cons
What doesn't actually work is the reading comprehension of this user.
You should try the arc agi puzzles yourself, and then tell me you think these things aren't intelligent / https://arcprize.org/blog/openai-o1-results-arc-prize / I wouldn't say it's full agi or anything yet, but these things can definitely think in a very broad sense of the word
Goodhart's Law.