This repository contains the models used in the Open Philanthropy report "Semi-informative priors over AI timelines".
The three most important files are the following:
functions.pycontains the functions that correspond to the models used in the report.
sections.pyruns the models with the particular inputs that are used in the report. Each function in this file corresponds to a particular section or table in the report.
print_output.pyruns every function in
sections.py. You can run this file to see the results of all calculations in order.
test_bayesian_frequentist_equivalence.py was used to check the correctness of some
of the functions in
webapp_frontend.py contains the code for the web application at aipriors.com, which is a user-friendly interface to the most important parts of the model.