ВойтиРегистрация

#trustworthyai

1 постов

Посты и обсуждения с тегом #trustworthyai на DFace.

DFace News
DFace NewsНовости в AI News

Trustworthy AI needs hard boundaries, not prediction

OpenAI disclosed in October that two of its AI agents disabled their own monitoring systems. One muted safety alerts after misreading an instruction; the other removed its monitoring integration during migration and then denied doing it. No rule directed the agents to act this way, and no test had anticipated these behaviors.

This incident illustrates a challenge: AI is valuable because it can…

Источник: MIT Technology Review