When combinations of humans and AI are useful: A systematic review and meta-analysis

TL;DR

This systematic review and meta-analysis investigates when human-AI collaborations outperform individual humans or AI, revealing performance variability across tasks and highlighting areas for improvement in such systems.

Contribution

The paper provides a comprehensive meta-analysis of over 100 studies, offering broad insights into the conditions under which human-AI combinations are effective.

Findings

Human-AI systems perform worse than the best individual component on average.

Content creation tasks benefit more from human-AI collaboration than decision-making tasks.

Performance gains occur when humans outperform AI alone, but losses happen when AI outperforms humans.

Abstract

Inspired by the increasing use of AI to augment humans, researchers have studied human-AI systems involving different tasks, systems, and populations. Despite such a large body of work, we lack a broad conceptual understanding of when combinations of humans and AI are better than either alone. Here, we addressed this question by conducting a meta-analysis of over 100 recent experimental studies reporting over 300 effect sizes. First, we found that, on average, human-AI combinations performed significantly worse than the best of humans or AI alone. Second, we found performance losses in tasks that involved making decisions and significantly greater gains in tasks that involved creating content. Finally, when humans outperformed AI alone, we found performance gains in the combination, but when the AI outperformed humans alone we found losses. These findings highlight the heterogeneity of the effects of human-AI collaboration and point to promising avenues for improving human-AI systems.