Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

this appears to be a(n incorrect!) gpt comment


No, it's a carefully constructed logical argument. Miles ahead of anything GPT is capable of producing, based on my experiments with it.

There's a clear signal that it's constructed by a person with advanced mathematical training: the laconic, "high points only" style. It makes small, but nontrivial and correct, logical leaps, expecting that you will work through the details needed to justify the leap yourself.

For example, the reasoning in this sibling comment is needed to justify the "n divides m" claim https://news.ycombinator.com/item?id=35638344

In my experience, GPT writes much more verbose arguments and cannot reliably reason at this level.


But if this were a college homework assignment you would be expected to explain briefly why n divides m, the comment did skip over it which can confuse people.


What drove you to this conclusion? We have an enlightening reply explaining how we know it isn't, but I would also like to know what heuristic you used (so that we know to be wary of it!)


Thanks for this awareness exercise. Still, I forget to add gpt to my internal reading filter




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: