# When Can Human-AI Teams Outperform Individuals? Tight Bounds with Impossibility Guarantees

arXiv: https://arxiv.org/abs/2605.08710 · HTML: https://mlsanity.com/paper/2605.08710/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 8 of its 32 references that are in the map.
- First posted on arXiv 9 May 2026.

## Builds on (most cited first, 8 of 8)

- [Does the Whole Exceed its Parts? The Effect of AI Explanations on Complementary Team Performance](https://mlsanity.com/paper/2006.14779/) (2020)
- [Effect of Confidence and Explanation on Accuracy and Trust Calibration in AI-Assisted Decision Making](https://mlsanity.com/paper/2001.02114/) (2020)
- [Consistent Estimators for Learning to Defer to an Expert](https://mlsanity.com/paper/2006.01862/) (2020)
- [On a Formal Model of Safe and Scalable Self-driving Cars](https://mlsanity.com/paper/1708.06374/) (2017)
- [When combinations of humans and AI are useful: A systematic review and meta-analysis](https://mlsanity.com/paper/2405.06087/) (2024)
- [A Unified Theory of Diversity in Ensemble Learning](https://mlsanity.com/paper/2301.03962/) (2023)
- [Who Should Predict? Exact Algorithms For Learning to Defer to Humans](https://mlsanity.com/paper/2301.06197/) (2023)
- [Complementarity in Human-AI Collaboration: Concept, Sources, and Evidence](https://mlsanity.com/paper/2404.00029/) (2024)

## Abstract (arXiv)

Human-AI teams fail to outperform their best member in 70% of studies, yet no theory specifies when complementarity is achievable. We derive tight bounds for the broad class of confidence-based aggregation rules by integrating signal detection theory with information-theoretic analysis, yielding four results: (1) a complementarity theorem (teams outperform individuals iff error correlation ρHM < ρ^*, with ρ^* ≈ a in the symmetric near-chance regime); (2) minimax bounds showing gains scale as Theta(√Delta d) with metacognitive sensitivity difference; (3) an impossibility result proving no confidence-based aggregation rule achieves complementarity when ρHM ≥ ρ^*; and (4) multi-class generalization ρ^*K ≈ ρ^*/√K-1. Predictions match observed team accuracy (R = 0.94 on ImageNet-16H, R = 0.91 on CIFAR-10H) and the multi-class threshold scaling holds on human data (R = 0.93, K = 16), with robustness under non-Gaussian distributions. The framework explains why complementarity is rare and provides actionable design formulas; results apply to aggregation, not to interactive deliberation that generates novel answers.