Igor delves into the political dimension of AI risk, highlighting how control over outcomes intersects with algorithmic procedures. He questions who holds the power to influence outcomes and whether AI alignment involves aligning preferences in a multi-agent setting.
