Topic: state of ai/safety
Context: Said by OpenAI’s head of alignment research in the week his company disclosed six incidents. A judgement about the industry’s readiness, including his own employer’s, rather than a checkable proposition.
Why this score? OpenAI's head of alignment research. Authoritative about his own team's findings and his company's practices, and the remarks recorded here are partly against his employer's commercial interest; but he speaks for a lab describing its own safety record, which is a disclosed stake. 0.7.