According to a proposal shared on LessWrong, researcher Dawn Drescher identifies a gap in AI governance planning: most thinking happens at the strategic level—weighing big risks, understanding timelines, debating policy approaches. But there's little work on the operational and tactical level: what policies actually look like when they meet real institutions and changing circumstances. Drescher proposes using computational exploratory modeling—specifically methods like Robust Decision Making—to test governance approaches against thousands of possible futures rather than betting on a few predictions. Rather than trying to predict the most likely scenario, you design policies that stay effective even when reality surprises you. She suggests packaging this as software, possibly paired with consulting services that adapt the models to each think tank's specific situation.

Listen to this story

Hear this and more stories in a personalized audio briefing.

Open The Chonkerton