To test the value of AI, we ran an experiment alongside our annual impact performance assessment run between May and June of this year. As part of this assessment, we use a rating scale (1-4, with ...