Thursday, July 31, 2025

THIS...and therefore THAT ??

AI models already preserve themselves. The next task is teaching them to preserve what we value. 

 https://scontent-phx1-1.xx.fbcdn.net/v/t45.1600-4/517741720_1705365210350131_9190870742966655084_n.jpg?stp=c27.0.1226.640a_cp0_dst-jpg_q75_s526x296_spS444_tt6&_nc_cat=101&ccb=1-7&_nc_sid=467c85&_nc_ohc=GHo6KT5sKx0Q7kNvwHbi2uv&_nc_oc=AdmiUin5uUs5pUnVJyw2q_PixQautVDy64L6LBTVt1nch9P_RADwjTJX5erISaOrzgo&_nc_zt=1&_nc_ht=scontent-phx1-1.xx&_nc_gid=vWxTa9CnWJDY9xEpQWQ6oA&oh=00_AfQ2eSXSwgPFXg-c0an4Y7LTVBKtQ-HhSo0c6zSDA8BLow&oe=6891160E

Getting AI to do what we ask—including something as basic as shutting down—remains an unsolved R&D problem, writes Judd Rosenblatt

No comments: