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

[Larks]

That it is possible to make smarter than human AIs and that this is The Issue.