[anoncow]

Can we arrive at some absolute rules about AI using the interactions?

[simonh]

At best you would get rules about this particular implementation and training approach. I don't see why we would expect those to apply to AI generally.