He thinks the answers which actually were, were, and the ones which actually were not, were not The idea that no humans were involved in altering either the mo

He thinks the answers which actually were, were, and the ones which actually were not, were not The idea that no humans were involved in altering either the model or the API, shortly after the model received bad publicity, and the alterations were specifically to the questions asked, in the way they were asked, and did not apply to the same question asked a different way, seems pretty unlikely