Reciprocal Human-AI Collaboration: Designing Configuration and Delegation for Continual Learning

Solving problems by human-AI configurations will likely become a pervasive practice. Traditional models of delegating tasks between humans and machines must be revisited in light of the differences in the learning of humans versus intelligent machines; performance can no longer be the sole criterion for task allocation. We propose a new human-AI configuration called a reciprocal human-machine learning (RHML) configuration and offer a new procedure for delegating tasks dynamically that begins with determining the desired level of machine autonomy.