My perception is that your job is to build a model to psychologically profile people at scale to manipulate them into buying stuff they wouldn't buy otherwise. Now, first of all, that seems incredibly unsettling to me in its own right already.

But where it gets really unsettling is if you think even a little bit past the immediate usecase: Once you have this sort of model, who knows what sort of other stuff you can manipulate people into? Should people like Zuckerberg really have the power to do this? And if you trust Zuckerberg, what about other players? Say an authoritarian state somehow managed to get control over these models. Doesn't it seem like you're building an incredibly dangerous thing here?

Wonder how you feel about these things and if you've ever had similar concerns. Like, I get that there's probably all sorts of interesting engineering problems involved. But one could say the same thing about von Braun building the V2 - these things aren't just puzzles in isolation!