Very interesting read! Your piece made me ponder upon a level deeper where we might observe and experience the concepts of “utilitarianism” with AI agents. This could potentially entail agents’ misdemeanour against humans who fed them incorrect or invalid data over the course of the AGI phase. Definitely a long-term thought but something that’s not impossible to rule out in today’s time!