Geek | junior developer vs machine learning

According to a new study from Princeton on ChatGPT-4 for real-world coding: Only 1.7% of the time was a solution generated that worked