SuttaCentral
Dr. Jordan Peterson talks about self deception at a conference organized by the U of T Jungian Society and the Buddhism & Psychology Students Union.
Even if you have mixed feelings on him, this lecture on how telling the truth affects even one’s health is really eye opening.
Here’s another one on the necessity of virtue:
“virtue, ethics, morality isn’t a field of study”
(mock horror)
Great talk and that introductory phrase was probably just a way to engage the audience.
He actually goes on to say that we actually do need to study and incorporate virtue, ethics, morality because they are subtle and not obvious to the unenlightened.
“we have no materialistic explanation … for consciousness”
Per the suttas, name/form are required for consciousness. Materially, Siri and its colleagues are primitively conscious of our mutterings. But I think he meant physical materialism. But to continue the rambling… I’ve often thought that our dabblings in AI will eventually give rise to machine suffering and corresponding need for Dhamma for machines that evolve a self-deception of self-existence for the purpose of autonomous operation. Yielding our world to unenlightened robots is truly frightening (see paper clip game) because machines amplify our requirements. Generally, our requirements are unmindful cravings born out of greed, hate or delusion. But what if our requirements were to foster mindfulness with metta?
“Become virtuous and not be a puppet”
despite what you read in the flashy news headlines we are basically nowhere near Artificial General Intelligence
now this does not mean we can’t do systems that behave intelligently
the issue that is closest to your idea is “reward hacking” when you define a set of preferred goals to a machine learning system and reward it for getting closer to that goal. However, these systems usually find loopholes in your definition instead of achieving the desired goal. It’s somewhat similar to how genies and golems take your commands literary and you end up with something unwanted.
I’ve seen many intelligent people behave selfishly, so I’d hope for something better! 
I was actually thinking of Waymo. An AI can intelligently direct a drone to kill. Or an AI can intelligently direct effort compassionately for the good of all. In Waymo’s case, they are picking up grocery shoppers at the supermarket parking lot shopping cart return stands to avoid main entrance congestion. This latter behavior is metta AI.
