tailcalled answers Are there any impossibility theorems for strong and safe AI?