Noah Scales comments on Posit: Most AI safety people should work on alignment/​safety challenges for AI tools that already have users (Stable Diffusion, GPT)