It blows my mind how bad the standard of basic logical arguments are on lesswrong, which is supposed to be a rationalist website, and has the sort of smug name that would challenge me to hold myself to a higher standard than this.

Here we have an article which amounts to “I have used AI to do some unimpressive research. I have read some unimpressive AI-assisted research. So AIs are unimpressive at research.”

Extrapolate from a tiny sample and establish a universal conclusion. Then look yourself in the mirror and congratulate yourself for how much “less wrong” you are than all the irrational schmucks out there.