Axios reports over 60 MAGA allies sent President Trump a letter urging a humans first policy that requires vetting of AI systems prior to public release
Signature review found repeated entries from Humans First and 39 pastors.
The gov't approval part has concentration of power issues (not conservative!); the focus on pre-deployment vs. the full lifecycle is a relic of a few years ago; private orgs should have a role in testing, not just, gov't, for concentration of power + expertise reasons; etc.
Miles Brundage@Miles_BrundageI'm glad to see conservative groups more engaged on AI safety lately in sometimes helpful ways, but I don't think this is a good / helpful policy position
I also worry about liberals trying to outdo "conservative" ideas like this with even more harmful ideas
We should get the basics in place before considering more extreme things - recall we don't even have a federal transparency requirement yet, or a well-funded CAISI, or (etc.)
Miles Brundage@Miles_BrundageThe gov't approval part has concentration of power issues (not conservative!); the focus on pre-deployment vs. the full lifecycle is a relic of a few years ago; private orgs should have a role in testing, not just, gov't, for concentration of power + expertise reasons; etc.