A machine's goal should be to maximize human values, even if it is uncertain what those values are.
From “3 principles for creating safer AI | Stuart Russell” — TED
see it said at 7:05
Did you know?
Surprising facts — each proven at the exact second of its video