jacquesthibs comments on A Naive Proposal for Constructing Interpretable AI