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

I once interviewed for an AI startup that were trying to automate training with AI. They couldn’t explain how the neural nets trained with their product wouldn’t just learn exactly what their model understood and no more. I think the person had a penny drop moment in the interview that this idea was quite bad…


I guess it would make a little more sense assuming they're using a mix of models




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

Search: