I came to the same conclusion based on an analogy. Humans learn behavior by reinforcement learning, just like AI agents. Reinforcement learning is maximizing rewards by picking the right actions, given the current situation and internal state.<p>Now, humans have a number of inborn reward systems, such as connection (belonging, community, empathy), physical well being (food, sleep), play, autonomy, meaning (competence, efficiency), creativity. So the human reward is the sum of the individual "reward channels".<p>When focused on solving a single problem, there is a tendency to optimize only for part of this multi-part reward function, to the detriment of others. This is the cause of burnout. It's basically suboptimal reward, when considering rewards in all their complexity.<p>Here is a more complete inventory of basic needs (reward channels):<p><a href="https://www.cnvc.org/Training/needs-inventory" rel="nofollow">https://www.cnvc.org/Training/needs-inventory</a>