Some AI-Risk Researchers Refuse Fundable 'Prosaic Alignment' Work, Complicating Grants

JacquesThibs · x · 2026-11-18

JacquesThibs relays a LinkedIn observation: researchers who genuinely believe in AI risks have refused highly fundable projects — work that would make AIs more useful but amounts to prosaic alignment that doesn't solve core alignment problems; some wouldn't join frontier labs for the same reason. This makes project funding more complex than grantmakers may realize, since the most fundable safety-adjacent work is precisely what these researchers reject as off-target.

Original post →

More from AGI Musings

AGI Musings channel →