# Align When They Want, Complement When They Need! Human-Centered Ensembles for Adaptive Human-AI Collaboration

arXiv: https://arxiv.org/abs/2602.20104 · HTML: https://mlsanity.com/paper/2602.20104/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 20 of its 41 references that are in the map.
- First posted on arXiv 23 Feb 2026.

## Builds on (most cited first, 20 of 20)

- [Training language models to follow instructions with human feedback](https://mlsanity.com/paper/2203.02155/) (2022)
- [Outrageously Large Neural Networks: The Sparsely-Gated Mixture-of-Experts Layer](https://mlsanity.com/paper/1701.06538/) (2017)
- [Gradient Surgery for Multi-Task Learning](https://mlsanity.com/paper/2001.06782/) (2020)
- [A General Language Assistant as a Laboratory for Alignment](https://mlsanity.com/paper/2112.00861/) (2021)
- [AdapterFusion: Non-Destructive Task Composition for Transfer Learning](https://mlsanity.com/paper/2005.00247/) (2020)
- [Consistent Estimators for Learning to Defer to an Expert](https://mlsanity.com/paper/2006.01862/) (2020)
- [Predict Responsibly: Improving Fairness and Accuracy by Learning to Defer](https://mlsanity.com/paper/1711.06664/) (2017)
- [Learning to Complement Humans](https://mlsanity.com/paper/2005.00582/) (2020)
- [MaxMin-RLHF: Alignment with Diverse Human Preferences](https://mlsanity.com/paper/2402.08925/) (2024)
- [What Large Language Models Know and What People Think They Know](https://mlsanity.com/paper/2401.13835/) (2024)
- [Controllable Preference Optimization: Toward Controllable Multi-Objective Alignment](https://mlsanity.com/paper/2402.19085/) (2024)
- [Appropriate Reliance on AI Advice: Conceptualization and the Effect of Explanations](https://mlsanity.com/paper/2302.02187/) (2023)
- [Human-AI Collaboration via Conditional Delegation: A Case Study of Content Moderation](https://mlsanity.com/paper/2204.11788/) (2022)
- [Do Large Language Models Perform the Way People Expect? Measuring the Human Generalization Function](https://mlsanity.com/paper/2406.01382/) (2024)
- [MAP: Multi-Human-Value Alignment Palette](https://mlsanity.com/paper/2410.19198/) (2024)
- [Designing for human-AI complementarity in K-12 education](https://mlsanity.com/paper/2104.01266/) (2021)
- [Does More Advice Help? The Effects of Second Opinions in AI-Assisted Decision Making](https://mlsanity.com/paper/2401.07058/) (2024)
- [Taking Advice from (Dis)Similar Machines: The Impact of Human-Machine Similarity on Machine-Assisted Decision-Making](https://mlsanity.com/paper/2209.03821/) (2022)
- [Fast Feature Selection with Fairness Constraints](https://mlsanity.com/paper/2202.13718/) (2022)
- [Human-AI Collaboration: Trade-offs Between Performance and Preferences](https://mlsanity.com/paper/2503.00248/) (2025)

## Built on by (most cited first, 1 of 1)

- [Addressing the Synergy Gap: The Six Elements of the Design Space](https://mlsanity.com/paper/2605.21635/) (2026)

## Abstract (arXiv)

In human-AI decision making, designing AI that complements human expertise has been a natural strategy to enhance human-AI collaboration, yet it often comes at the cost of decreased AI performance in areas of human strengths. This can inadvertently erode human trust and cause them to ignore AI advice precisely when it is most needed. Conversely, an aligned AI fosters trust yet risks reinforcing suboptimal human behavior and lowering human-AI team performance. In this paper, we start by identifying this fundamental tension between performance-boosting (i.e., complementarity) and trust-building (i.e., alignment) as an inherent limitation of the traditional approach for training a single AI model to assist human decision making. To overcome this, we introduce a novel human-centered adaptive AI ensemble that strategically toggles between two specialist AI models - the aligned model and the complementary model - based on contextual cues, using an elegantly simple yet provably near-optimal Rational Routing Shortcut mechanism. Comprehensive theoretical analyses elucidate why the adaptive AI ensemble is effective and when it yields maximum benefits. Moreover, experiments on both simulated and real-world data show that when humans are assisted by the adaptive AI ensemble in decision making, they can achieve significantly higher performance than when they are assisted by single AI models that are trained to either optimize for their independent performance or even the human-AI team performance.