Self-programming inductive machines have enough creativity to recognise the truth of Gödel sentences—and to beat the Lucas argument.