Balancing Performance and Human Autonomy With Implicit Guidance Agent

The human-agent team, which is a problem in humans and autonomous agents collaborate to achieve one task, typical human-AI collaboration. For effective collaboration, want have an plan, but realistic situations, they might difficulty calculating the best plan due cognitive limitations. In this case, guidance from agent that has many computational resources may be useful. However, if guides human behavior explicitly, feel lost autonomy are being controlled by agent. We therefore investigated implicit offered means of agent’s behavior. With type guidance, acts way makes it easy for find collaborative can then improve plan. Since improves their voluntarily, he or she maintains autonomy. modeled with integrating Bayesian Theory Mind into existing collaborative-planning algorithms demonstrated through behavioral experiment enabling maintain balance between improving plans retaining