Hacker News new | past | comments | ask | show | jobs | submit login

AI has a ton of impossibility results. the "no free lunch theorems" are probably the best known. people now actively think in terms of designing assumptions, priors, or inductive biases to make their systems work better on the data they actually care about, because there's typically no solution that is always superior without tradeoffs.

https://en.wikipedia.org/wiki/No_free_lunch_in_search_and_op...




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

Search: