Autonomous artificial intelligence and liability: a comment on List

Christian List argues that responsibility gaps created by viewing artificial intelligence (AI) as intentional agents are problematic enough that regulators should only permit the use of autonomous AI in high-stakes settings where AI is designed to be moral or a liability transfer agreement will fill any gaps. This work challenges List’s proposed condition. A requirement for ‘moral’ AI is too onerous given technical challenges and other ways to check AI quality. Moreover, transfer agreements only plausibly fill responsibility gaps by applying independently-justified group responsibility attribution norms such that AI raises no unique regulatory norms