This “shitty distraction”
doesn’t make things easier in the considerations about where intent lies. If the AI does not know what it means when it says things, then until it does, how is intent imputed in the law of contract? If this is the space we are operating in, then it strikes me that if AI enters into a contract, a court would have to take a giant leap to ascribe perptual intent in the exercise of launching such AI to all contractual activity the AI undertook later. It’s almost like saying that as long as I pass any glass of water under a bobbing bird thingmabob, I always intended that it should take a drink.