A No Free Lunch Theorem for Human-AI Collaboration
This provides theoretical guidance for designing human-AI collaboration systems, though it is incremental in extending no-free-lunch theorems to this domain.
The paper tackles the challenge of achieving complementarity in human-AI collaboration for binary classification, showing a 'No Free Lunch'-style result that any deterministic strategy not always deferring to the same agent can sometimes perform worse than the least accurate agent.
The gold standard in human-AI collaboration is complementarity -- when combined performance exceeds both the human and algorithm alone. We investigate this challenge in binary classification settings where the goal is to maximize 0-1 accuracy. Given two or more agents who can make calibrated probabilistic predictions, we show a "No Free Lunch"-style result. Any deterministic collaboration strategy (a function mapping calibrated probabilities into binary classifications) that does not essentially always defer to the same agent will sometimes perform worse than the least accurate agent. In other words, complementarity cannot be achieved "for free." The result does suggest one model of collaboration with guarantees, where one agent identifies "obvious" errors of the other agent. We also use the result to understand the necessary conditions enabling the success of other collaboration techniques, providing guidance to human-AI collaboration.