[assemblyman]

Even with thick desires, I sometimes find myself day-dreaming about the state when I have mastered a skill or understood a topic deeply. At the same time, I know from experience that the process never ends. Even when one does master a skill, one is deeply aware of what one doesn't know or understand or what one is not good at within that domain.

What helps me is to focus on today. If I can spend even an hour on a topic and get lost in it or even get frustrated by it, it is time well-spent. I was going to say "it is progress" instead of "time well-spent" but even that's a trap. Progress implies moving forward in a preferred direction. While I can't say I don't want to make progress, I am training myself to care less about it. It is really the time spent engaging that's most valuable (at least to me).

[neom]

You might find this interesting: https://en.wikipedia.org/wiki/Up%C4%81d%C4%81na and https://www.wisdomlib.org/definition/bhavatanha

[nrhrjrjrjtntbt]

Oh yeah decades in I still feel I know f-all about programming. Doesn't help the field keeps expanding expintentially. E.g. I look most things up. I am basicially a slow LLM!

[bigfishrunning]

You're kind of the opposite of a slow LLM. LLMs don't look anything up, they enthusiastically assert that they're correct. They have no desire to know anything.

[CamperBob2]

LLMs don't look anything up, they enthusiastically assert that they're correct.

Says someone who lectures on how LLMs worked two years ago.

[bigfishrunning]

https://openai.com/index/why-language-models-hallucinate/ https://cdn.openai.com/gpt-5-system-card.pdf

according to openai, the least likely model to hallucinate is gpt-5-thinking-mini, and it hallucinates 26% of the time. Seems to me the problems of LLMs boldly producing lies are far from solved. But sure, they lied years ago too.

[CamperBob2]

according to openai, the least likely model to hallucinate is gpt-5-thinking-mini, and it hallucinates 26% of the time.

You're not so bad at hallucinating, yourself. We find that gpt-5-main has a hallucination rate (i.e., percentage of factual claims that contain minor or major errors) 26% smaller than GPT-4o ...

That's the only reference to "26%" that I see in the model card.

[bigfishrunning]

I get that in the age of AI, you didn't want to read the data i linked; that's fine. your ctrl-f search found a reference to 26%. However, on page thirteen, the rate is described as 0.26; I interpreted that as 26% because it's cross referenced in the blog post that i also linked.

[godelski]

If you didn't daydream like that would you have the motivation to pursue it? Are not those daydreams your kind encouraging you? "Look how great it'll be, this is why you'll put in the hard work now". You can get trapped in the dreams, of course, but they're useful too

[dtjohnnyb]

David Epstein calls this "desirable difficulty" in the book Range.

Interestingly he recently discussed how using LLMs tends to remove this desirable difficulty: https://davidepstein.substack.com/p/a-risk-of-cognitive-conv...

This means that the results (both of the task and of the learning by the student) are lower if the student uses an LLM first, but slightly improves if they use it second