Are you sure? We're already getting stories over here in the West where people are struggling with the results AI gives them when applied to people because the results are basically Wrongthink, and are trying to figure out how to get the AIs to conform better to their ideology and preconceived notions of what the AIs should be outputting. It's not a terribly far trip from there to Deutsche Physik or Lysenkoism, except wrapped in a much denser fog of complex math and enough complicated details to construct any narrative you want. Using AI to ideologically crack down on a population will eventually reach even the AI researchers and destroy their ability to maintain and develop the AI.

(I don't think it's an inevitable trip from struggling with AI outputs and preconceived ideological notions to Lysenkoism. But I think it is an extremely tempting path for the society as a whole. There are a lot of forces that push in that direction, as the nature of the AI and what they do and control moves from a scientific and mathematical matter to a political one.)