<ul data-eligibleForWebStory="true">A study conducted by multiple institutions outlined the use of AI to write code globally and its economic impact.The study trained a neural classifier to detect AI-generated Python functions in 80 million GitHub commits by 200,000 developers over six years.AI was estimated to have written 30.1% of Python functions by US contributors, with varying percentages in other countries.Newer GitHub users tend to use more AI than veteran users, and using AI 30% of the time can raise quarterly commits by 2.4%.Combining wage data with programming estimates, the study calculates that AI generates $9.6-14.4 billion annually in the US software sector.Researchers ensured the neural classifier correctly identifies AI-generated Python functions by collecting human-written functions and synthetic AI-written functions.They used GraphCodeBERT, a cutting-edge model, to tokenize and embed the code for classification, achieving a high detection precision.The study explored AI's impact on coding practices and economic implications in various countries and sectors.