@jeffjarvis love your discussions on AGI on twit. What do you think of the idea we won’t be able to create AGI until we can devise a reward/punishment system for a computer the way we are wired to produce reward/punishments for ourselves (dopamine, serotonin, adrenaline etc) and what would that system look like. Right now an algorithm has no benefit to improve itself, but if we could design something that gives it a dopamine like hit, it would want to.