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

Even Multivac said "Insufficient data for a meaningful answer" for some particular question. But the problem of all "safe" AIs is more related to HAL9000, fuzzy directives about what to do and not to based on subjective and incomplete rules that may lead to inappropriate answers.


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

Search: