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

imo, working memory at scale via improvements to more obscure methods like a DNC is what will be the next leap and make todays 'AI' seem mundane. Most AI today like deepfakes involve giving the model all memories available via input.


Scaling SHRDLU in a similar manner would also rock, though.




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

Search: