As I mentioned, if it evolves to the direction of authoritarian intelligence — in which the data collection is not consensual, in which the application of data (including its bias) has no way to appeal or at least compares badly to the judicial way of getting appeals done and so on — then of course it’s quite worrying. But the fact that we’re talking about this means that people already worry about it sufficiently — that people are working on tackling this very problem to make what I call “assistive intelligence” or “augmented collective intelligence” the predominant norm, at least in pro-democratic countries. Like the Global Partnership on Artificial Intelligence, focusing the AI research on that particular value system.