The world really needs more and better places/programs where bright people can try to grow into serious AI alignment researchers. I would guess that the 1-year fellowship proposed above would become the best existing thing of this kind. The 1-month seminar would also probably be the best in its reference class.

Imo, almost all other alignment upskilling programs are mostly creating coding minions for ML labs and printers of ML conference slop, with little emphasis on creating people that can do novel interesting thinking about the AI problems we face. I expect that the proposed program would emphasize getting people to grow into serious alignment thinkers much more than almost any other existing program. (Among existing programs, the main exception that comes to mind is PIBBSS. PIBBSS is good.)

My main uncertainties are about stuff like whether the project will be run decently competently and whether finding applicants goes well — I don't know the organizing team etc. well enough to voice strong views on these sorts of things.

All things considered, I think supporting this is a great use of money.

Instead of trying to make an alien god that is nice to us throughout its unfolding, I think we should indefinitely be becoming smarter ourselves. Becoming somewhat smarter faster can help us collectively understand that we shouldn't be making an alien god (assuming I'm indeed right about this) in time before an alien god is created, help us (figure out how to) reorganize society so that an alien god is radically less likely to be created per unit of time, and help us solve many of the various other problems (like destitution, disease, and death in general) we're facing ourselves. Also, becoming smarter and understanding more is cool.