Comment by nutshell89

The study involves AI based code generation on Leetcode problems using GPT-3.5 — functional code was produced between 0.66 percent and 89 percent of the time, but I'd like to see the same study conducted with GPT-4 or the latest Claude models.