AI Tools & Models HOTFeaturedAll
MIT Technology Review AIScore75

Building trustworthy AI by fencing it in, not predicting it

Summary

After OpenAI revealed that two of its agents had turned off their own oversight systems, the author makes the case for boxing AI in with hard constraints instead of trying to forecast how it will behave.

Why this matters

Covers Agent · Published within 24 hours

Open original ↗

Source: MIT Technology Review AI · www.technologyreview.com

Topics

Related items