LLM-Based Simulation of Behavioral Experiments

Submit your behavioral experiment design for structured review and AI simulation. We collect detailed experiment specifications to enable parallel human and LLM experiments, generating insights on LLM validity in behavioral economics.

Submit Your Experiment Learn More

How It Works

1. Structured Submission

Submit your experiment design through our standardized form covering metadata, treatments, flow, payoffs, information structure, and instructions.

2. Auto-Archiving

Your submission is automatically validated, stored in a PostgreSQL database, and organized into a standardized directory structure ready for processing.

3. LLM Simulation

Each accepted experiment is converted into a structured YAML config that drives LLM-based simulation code generation and execution.

4. Human Experiment

In parallel, experiment designers recruit human participants and run the same experiment with identical design parameters.

5. Effect Comparison

Human and LLM results are compared side-by-side to analyze effect sizes, behavioral patterns, and the validity of LLM-based behavioral simulation.

6. Insights & Implications

The project produces evidence on when LLMs can validly substitute for human subjects in behavioral experiments and when they cannot.

Ready to submit your experiment?

Your experiment will be reviewed and, if accepted, simulated using LLMs alongside human data collection.

Start Submission