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

> doesn't seem that GPT is significantly better or worse than a human in this regard

Probably, but I think a human is much more likely to realize that they made a mistake and good-luck convincing GPT that it's wrong about something non trivial.



Well it's trained to simulate humans and that includes simulating the errors.


This is an exceptionally deep, interesting, and important statement. Is it true? Does GPT-3 occasionally make spelling errors for instance? Is the rate of spelling errors higher or lower than the average human? How about versus the average in the training data? How does the prompt affect the frequency of such errors. Ditto for other kinds of objectively measurable error.


In my experience it makes less errors if you prompt it to reason carefully step by step.


Opposite of my experience. If you talk it through each step you can back it into persistently claiming that 0^2=1.


I'm pretty sure they're talking about errors it makes as part of neutral generation, not how much of a pushover it is.




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

Search: