From the article abstract: "All experiments were done with safety precautions (e... | Hacker News

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

		wiz21c 7 months ago \| parent \| context \| favorite \| on: AI Improves at Improving Itself Using an Evolution... From the article abstract: "All experiments were done with safety precautions (e.g., sandboxing, human oversight)." Do the authors really believe "safety" is necessary, that is, there is a risk that somethign goes wrong ? What kind of risk ?

datameta 7 months ago [–]

From what I understand, alignment and interpretability were rewarded as part of the optimization function. I think it is prudent that we bake in these "guardrails" early on.

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