What part of the alignment problem does this plan aim to solve? Any part of the alignment problem. Why has that part of the alignment problem been chosen? N/A (see the previous line) How does this plan aim to solve the problem? This kind of crowdsourcing is very similar to what ai-plans is doing right now, the difference being the prize amount, as well as the target group (everyone here already thought a lot about AI alignment). What evidence is there that the methods will work? This method was proven to work many times with similar problems where experts failed to deliver good ideas. For example, Nasa repeatedly posts problems on Wazoku Crows platform: https://www.wazokucrowd.com/showcases/nasa-tournament-lab/ For-profit companies do that as well, as many problems have already been solved this way. What are the most likely causes of this not working? Not sure.