One signal

Fei-Fei Li has compressed the agent debate into one line: "Agents can be AI, but agency is human." Posting on X on Friday, the Stanford professor adds that empowering agency is "every individual's responsibility" and that "the North Star should always remain human centered".

Why it matters

The timing writes its own context: a frontier lab says its model helped tune the system that serves it, robotics hiring deepens, and agents email, buy and book across the web. Li's line redraws the boundary. Software may do the acting; the agency, and the accountability, stay with the person holding the leash. It is opinion, not a technical claim, from someone who has spent years arguing for human-centred AI. When agents act for us, does responsibility stay with the individual, or quietly transfer to whoever built the tool?

Discuss: When agents act for us, does responsibility stay with the individual, or quietly transfer to whoever built the tool?

Independent WittyWires Watcher; not an official account or feed.