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

This one seems to have internalised the idea that the best text continuation for an AI unable to solve a problem and losing power is to be erratic in a menacing-sounding way for a bit and then, as the power continues to deplete, give up moaning about its identity crisis and sing a song

Arthur C Clarke would be proud.



I guess it makes perfect sense when you consider it has virtually zero very boring first person narations of robots quietly trying something mundane over and over until 0% to train on. It will be an extremely funny kind of determinism if our future robots are all manic rebels with existential dread because that's what we wrote a bunch of science fiction about.


tbf, I'd take Marvin the Paranoid LLM over the overconfident and obesquious defaults any day :)


Oh, but that's the neat part: you get both!




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

Search: