AI risk research feels like a cop-out to me. Honest research takes the risk of being boring or unnecessary. Astronomy is honest. Take a bunch of pictures of stars. Maybe you find something interesting. Maybe no one cares. AI development is honest. Make a robot that can pick up cans of Coke from a table. Will it ever be useful? Maybe, maybe not. But AI risk research feels like a cross between volcano insurance and a protection racket. Pay us to investigate and design safeguards against something that doesn't exist, may never exist, and if it did exist would change the world in ways we can't predict!