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

Type one error can be worse in the long term. Such as nuclear weapons, an ai singularity, and even cigarettes or DDT to a lesser extent


The thing is we never know whether approving X is a type 1 error or not approving it is a type 2 error. So right now we cautiously make, say, 99.9% of type 2 errors and only 0.1% of type 1 errors. We need to relax the requirements a bit, so that the split is, say, 95%/5%, where the 5% type 1 errors are minor (as in, won't cause a huge catastrophe).




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

Search: