# 2-Step Agent: A Framework for the Interaction of a Decision Maker with AI Decision Support

arXiv: https://arxiv.org/abs/2602.21889 · HTML: https://mlsanity.com/paper/2602.21889/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 14 of its 33 references that are in the map.
- First posted on arXiv 25 Feb 2026.

## Builds on (most cited first, 14 of 14)

- [Performative Prediction](https://mlsanity.com/paper/2002.06673/) (2020)
- [Hacking Smart Machines with Smarter Ones: How to Extract Meaningful Data from Machine Learning Classifiers](https://mlsanity.com/paper/1306.4447/) (2013)
- [Human-Algorithm Collaboration: Achieving Complementarity and Avoiding Unfairness](https://mlsanity.com/paper/2202.08821/) (2022)
- [A Decision Theoretic Framework for Measuring AI Reliance](https://mlsanity.com/paper/2401.15356/) (2024)
- [Eavesdrop the Composition Proportion of Training Labels in Federated Learning](https://mlsanity.com/paper/1910.06044/) (2019)
- [Towards Human-AI Complementarity with Prediction Sets](https://mlsanity.com/paper/2405.17544/) (2024)
- [Experimental Evaluation of Algorithm-Assisted Human Decision-Making: Application to Pretrial Public Safety Assessment](https://mlsanity.com/paper/2012.02845/) (2020)
- [Human Expertise in Algorithmic Prediction](https://mlsanity.com/paper/2402.00793/) (2024)
- [Optimal regimes for algorithm-assisted human decision-making](https://mlsanity.com/paper/2203.03020/) (2022)
- [Controlling Counterfactual Harm in Decision Support Systems Based on Prediction Sets](https://mlsanity.com/paper/2406.06671/) (2024)
- [Designing Algorithmic Recommendations to Achieve Human-AI Complementarity](https://mlsanity.com/paper/2405.01484/) (2024)
- [The risks of risk assessment: causal blind spots when using prediction models for treatment decisions](https://mlsanity.com/paper/2402.17366/) (2024)
- [Evaluating and Correcting Performative Effects of Decision Support Systems via Causal Domain Shift](https://mlsanity.com/paper/2403.00886/) (2024)
- [Bayesian Inference of Training Dataset Membership](https://mlsanity.com/paper/2506.00701/) (2025)

## Abstract (arXiv)

Predictions from ML models support human decision making in several fields, including high-stakes ones such as healthcare and the judiciary. Yet, we still lack a clear understanding of how decision makers learn from ML-based decision support (ML-DS). In this paper, we introduce a general computational framework, the 2-Step Agent, to capture this process. As a prediction from an ML model contains information about the training data, a prediction can also be used for inference. Our framework models (i) how a prediction for a new observation affects the beliefs of a rational Bayesian agent, and (ii) how this change in beliefs affects the estimation of causal effect, the downstream decision, and the subsequent outcome. In addition to the framework itself, we make three contributions. First, for the linear Gaussian setting, we derive a tractable solution for the challenging Bayesian inference problem we introduced, i.e. one in which the agent infers from an ML prediction. Second, we experimentally identify conditions under which ML-DS is beneficial. Third, we show that a single misaligned prior belief can be sufficient for ML-DS to lead to worse downstream outcomes compared to no decision support even when the ML model is well-specified and the agent is perfectly rational. Hence, even under ideal conditions, ML-DS can do more harm than good.