Evaluating the feasibility of SI's plan — AI Alignment Forum