AI Safety Debate Distinguishes Desired Laws From Retroactive Punishment
Twitter replies discuss desired laws versus ex post facto judgments for AI developers.
Rob Miles argued that statements about punishing AI norm-breakers mean enacting future laws in a just world, not retroactive jail time for past legal acts. Matthew Barnett replied that vaccine developers also broke norms and interpreted similar tweets as calling for actual ex post facto judgment, referencing generated headlines on Nuremberg trials for AI safety issues. The cluster centers on this exchange about accountability norms in AI development.
Combined views
86
2 posts, first seen 2h ago
AI Safety Debate Distinguishes Desired Laws From Retroactive Punishment
Twitter replies discuss desired laws versus ex post facto judgments for AI developers.
Rob Miles argued that statements about punishing AI norm-breakers mean enacting future laws in a just world, not retroactive jail time for past legal acts. Matthew Barnett replied that vaccine developers also broke norms and interpreted similar tweets as calling for actual ex post facto judgment, referencing generated headlines on Nuremberg trials for AI safety issues. The cluster centers on this exchange about accountability norms in AI development.
Reactions from ranked influencers
2 posts
@MatthewJBar @sharongoldman @ilex_ulmus They've certainly broken existing norms, in her view. But in this case 'deserves' is important. Like "in a just world, we would have a law such that you'd be in jail for doing that", not "we should enact a retroactive law to jail you for something that was legal when you did it"
@robertskmiles @sharongoldman @ilex_ulmus One could equally argue that vaccine developers broke existing norms, in one's own view. Also, I don't interpret tweets like the one below as merely meaning that "in a just world, we'd punish this behavior". It seems more like she's referring to actual ex post facto judgement.