Richard_Kennaway comments on Patternist friendly AI risk