The use of AI requires a proactive approach to ensure its outcome leads to good, not bad. Observing human behavior can aid in this, as some people may act irrationally even when presented with guarantees.