AGIs, from a systems point of view, are a part of the Sapiens plurum group that invented them. As a cognitive component of a larger system, they need not have innate drives. They could potentially respond to alerts and seek homeostasis in other entities, such as humans. This would assure that AGIs act to benefit humans. This then reinforces the critical nature of the AGI's definition of ingroup and outgroup. Whose feedback is considered relevant? Who is part of We and who are They?
Despite the fact that this age-old dilemma remains, by deriving AGI motivations from human emotion, we insure that We does not become AGIs and They, humans.