If it is true that...

If we don't actively build AI for good, its evil outcomes may manifest

and

AI has the potential for evil outcomes

Then it must be true that...

We must actively build AI for good