# Amadeus: When Models of People Meet

arXiv: https://arxiv.org/abs/2609.35835 · HTML: https://mlsanity.com/paper/2609.35835/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 10 of its 16 references that are in the map.
- First posted on arXiv 24 Sep 2026.

## Builds on (most cited first, 10 of 10)

- [Generative Agents: Interactive Simulacra of Human Behavior](https://mlsanity.com/paper/2304.03442/) (2023)
- [SOTOPIA: Interactive Evaluation for Social Intelligence in Language Agents](https://mlsanity.com/paper/2310.11667/) (2023)
- [LLM Agents Grounded in Self-Reports Enable General-Purpose Simulation of Individuals](https://mlsanity.com/paper/2411.10109/) (2024)
- [Aligning Superhuman AI with Human Behavior: Chess as a Model System](https://mlsanity.com/paper/2006.01855/) (2020)
- [Detecting Individual Decision-Making Style: Exploring Behavioral Stylometry in Chess](https://mlsanity.com/paper/2208.01366/) (2022)
- [Learning to Imitate with Less: Efficient Individual Behavior Modeling in Chess](https://mlsanity.com/paper/2507.21488/) (2025)
- [Chessformer: A Unified Architecture for Chess Modeling](https://mlsanity.com/paper/2605.19091/) (2026)
- [Generative Modeling of Individual Behavior at Scale](https://mlsanity.com/paper/2502.14998/) (2025)
- [Toward Modeling Player-Specific Chess Behaviors](https://mlsanity.com/paper/2605.11893/) (2026)
- [Matilda: Engine-Agnostic Search with Human Policy Guidance](https://mlsanity.com/paper/2606.25176/) (2026)

## Abstract (arXiv)

With the constant advancements in AI, one possibility is to model agents after humans and, in turn, use these agents to carry out synthetic interactions. Such models could be used to predict interactions between their real counterparts, or potentially interactions at larger scales. In this paper, we test a more controlled version of this question through chess. We use 8 elite chess players, seal their direct pairwise games, learn each player independently using different methods, and then compose the resulting models on the withheld dyads. To evaluate the generated interactions, we use two measurements: opening-family total variation distance and win-draw-loss (WDL) total variation distance. M1 primarily improves WDL fidelity while producing smaller opening-family improvements, whereas M2 produces much larger opening-family improvements while having little effect on WDL-TV. For opening-family behaviour under M2, the correct assignment of the eight learned player identities also gives the closest match among all 8! = 40,320 possible assignments. These results show that at least some properties of previously unseen interactions can be recovered from independently learned individuals. The partial recovery observed here may reflect limitations of the current individual modelling methods rather than a fundamental limit on compositional interaction recovery. An additional post-hoc method that combines the two mechanisms improves both measurements, suggesting that recovery across these behavioural properties is not necessarily mutually exclusive.