G Gordon Worley III comments on Minimization of prediction error as a foundation for human values in AI alignment