AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
Why it matters
Argues that AI safety controls need to operate at both the model level (e.g. Scientist AI-style supervision) and the system level over AI agentAn AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.Full definition → scaffolds and harnesses, with independent verification and monitoring built in from the start rather than bolted on later.
