AI Incident Forecasting

Forecasting real-world AI failures before they happen

This site documents the AI Incident Forecasting project developed during the Apart AI Forecasting Hackathon.

We build statistical models on the AI Incident Database to:

  • Forecast future AI incident counts
  • Break forecasts down by risk domain and category
  • Quantify uncertainty using calibrated prediction intervals
  • Compare performance against naive baselines

Repository:
https://github.com/cluebbers/ai-incident-forecasting

Posts

subscribe via RSS