samwestby / humanaiteamsandci

Code for Collective Intelligence In Human-AI Teams: A Bayesian Theory of Mind Approach

Collective Intelligence In Human-AI Teams: A Bayesian Theory of Mind Approach

This is code for the following publication: Westby, S. & Riedl, C. (2023). Collective Intelligence in Human-AI Teams: A Bayesian Theory of Mind Approach. In Proceedings of the 37th AAAI Conference on Artificial Intelligence (2023). https://arxiv.org/abs/2208.11660

Workflow:

1. Download this directory
2. Change working directory to the downloaded HumanAITeamsAndCI directory
3. Create a Python 3.x environment and run pip install -r requirements.txt
4. Run python -m scripts/results_table to replicate Table 1
5. Run python -m scripts/figure_3 to replicate Figure 3
   Note: There was a problem with the random seeds so results and figures are marginally different
6. Run python -m scripts/grid_search to redo the parameter grid search
   • Each run requires approximately 50 GB
   • On an intel i7 with 8 GB of RAM, this takes approximately 40 minutes
   • Change lower and upper in main() to search a smaller parameter space
   • Run it twice. Once with self_actualize = True and another with self_actualize = False
7. Now we move on the the R code - replicating Figure 4 and the log likelihoods in Table 1
8. In your preferred R editor, run scripts/LikAnalysis_discovery.R
   • Line 3 and 76 require you to input preferred directories
   • This file aggregates the grid search data
   • Repeat to analyze both grid searches from above
9. Run rcode/LikAnalysis.R to generate the logLiks from Table 1
   • Input your preferred directories where noted
10. Run rcode/MessageToM.R to generate the plots in Figure 4
    • Input your preferred directories where noted