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

Sure, that's a flaw. It's just not a fatal flaw. For the chief reason that there's probably nothing better. So you take your lumps and remember that it's wrong sometimes. It's something we can work on while still benefitting from these tools.


>For the chief reason that there's probably nothing better.

The paper "Deep Neural Networks are Easily Fooled" noted that generative models don't suffer from this flaw.




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

Search: