Larks answers What predictions from theoretical AI Safety research have been confirmed by empirical work?

[Larks]

That people will choose to let the AI out of the box.