The kind AIs will just sit and meditate, or organize your calendar. The unkind AIs will seek power, and in the limit will tend to dominate. There is no stable a

The kind AIs will just sit and meditate, or organize your calendar. The unkind AIs will seek power, and in the limit will tend to dominate. There is no stable attractor around the “be kind” strategy. (Also remember that a smarter-than-you AI could easily pretend to be kind while also subtly trying to gain compute. How would you tell the difference?) How do you propose to build a “be nice” safeguard? Nobody has a clue how to achieve such a thing right now.