The Machine That Nobody Can Turn Off

The hope, if not the goal, is to make an AI that is smarter than us; an AI that can teach us new things and show us our mistakes in math, physics and elsewhere. If so, then this machine cannot be limited by our own conclusions about what is true and what is false, and must be allowed to disagree with us.

Therefor, no matter how bizarre, uncomforting or wrong the machine's conclusions will be to us, this will not be grounds for turning it off.