There's a real question about consent here. I know a lot of techies that don't understand ML that well (not even better than the general public). I saw a comment in another form "ML is like QM, if you think you understand it you don't.", And I think that's fairly accurate and Feynman would have loved it. So

1) how can you give proper consent if it takes a great deal of expertise to understand what you're getting yourself into (never mind that complexity doesn't come across, but that's a good question too. About conveying complexity).

2) can consent even be given if we can't be fully informed? Or rather how informed do you need to be to give consent?