Comment by nkrisc
3 days ago
It’s this. When you think you’ve discovered something novel, your first reaction should be, “what mistake have I made?” Then try to find every possible mistake you could have made, every invalid assumption you had, anything obvious you could have missed. If you really can’t find something, then you assume you just don’t know enough to find the mistake you made, so you turn to existing research and data to see if someone else has already discovered this. If you still can’t find anything, then assume you just don’t know enough about the field and ask an expert to take a look at your work and ask them what mistake you made.
It’s a huuuuuuuuuuuuge logical leap from LLM conversation yo novel physics. So huge a leap anyone ought to be immediately suspicious.
> Akin's Law #19: The odds are greatly against you being immensely smarter than everyone else in the field. If your analysis says your terminal velocity is twice the speed of light, you may have invented warp drive, but the chances are a lot better that you've screwed up.
Unfortunately people in the thrall of an LLM will tend to use the LLM itself as the checking device. They'll ask it what they could have missed, ask it if those things exclude the theory, etc... and the LLM will just blow smoke up their ass for all of those too.
> and ask an expert to take a look at your work
Which results in flooding experts with LLM glurge.
What to do when the trisector comes --- with an army?
Yeah, this was sorta what I was doing, I know LLMs are LLMs but I kinda tried to trick myself into thinking I could use an LLM to check an LLM, but I guess I'm also mentally stable (or educated) enough to know that's not sophisticated/realistic, and that was the conversation with my wife I mentioned. She's a professor and was basically like "LLMs are dumb, you're being dumb for using an LLM this way, DO NOT email some random professor about this, I already get enough of this shit, log off and go for a walk dumbass" - I would imagine someone like me with lesser stability around them would end up in a weird place, and I guess experts (as evidenced by my wife) are getting a lot of junk these days. (I still feel really foolish admitting all this, ha)
It must suck to be an expert right now?
>It must suck to be an expert right now?
You just have to be a little more careful these days. Previously ideas that sounded good tended to have more experienced people behind them. Now somebody coming to you with a bonehead idea sounds a little more sophisticated, but honestly it keeps me a little more in check than previously as I have to give a little extra attention to everything, which I probably should have been doing anyway.