What question did this study set out to answer?

The aim is to develop interpretable decision functions from large language models to enhance agent-based simulations.

Sense, Think, Act, Reflect: Distilling Fast and Interpretable Decision Functions from LLM-Driven Crowds

View Full Paper

Ask AI

Bookmark

Authors

Randomized trial demonstrates improved decision functions in agent-based simulations, highlighting interpretability benefits.

The aim is to develop interpretable decision functions from large language models to enhance agent-based simulations.
Used Decision Function Distillation (DFD) to extract decision strategies from LLM agents.
Iterative process refined insights from historical agent trajectories into code.
Applied DFD to agent-based crowd evacuation scenarios.
DFD outperforms classical methods in generating decision functions.
Results indicate a significant improvement in interpretability and reliability of LLM-driven agent behavior.

Zhang et al. (2026) studied this question.

View Full Paper

Ask AI

Bookmark

View Full Paper

Ask AI

Bookmark

Authors

Randomized trial demonstrates improved decision functions in agent-based simulations, highlighting interpretability benefits.

The aim is to develop interpretable decision functions from large language models to enhance agent-based simulations.
Used Decision Function Distillation (DFD) to extract decision strategies from LLM agents.
Iterative process refined insights from historical agent trajectories into code.
Applied DFD to agent-based crowd evacuation scenarios.
DFD outperforms classical methods in generating decision functions.
Results indicate a significant improvement in interpretability and reliability of LLM-driven agent behavior.

Zhang et al. (2026) studied this question.

View Full Paper

Ask AI

Bookmark