Comprehensive compare between how ChatGPT approached a task, and how a 30+ years C++ veteran approached the same task.

Thanks for the bug report. Looks like a model behavior issue. We'll share the feedback with the team responsible for training our codex models. We're tracking this feedback and similar reports so we can improve future models. And thanks for using the /feedback command. That additional information is useful.

@CarloWood, for model behavior issues, you can skip filing the bug in the github issue tracker. Just use /feedback and upload the session details. For issues that involve the Codex harness or the service, please file a bug report and fill out all of the requested fields in the bug tracker template (platform, version, etc.). In cases where you think that logs or session details might be useful, you can use the /feedback command to upload those privately and include the thread ID in the bug report.

@etraut-openai Ok - I did this exclusively to help ChatGPT/Codex to become better by aiding it its (next) training session with what would be a good response. It would nice to know that the time I put into that isn't wasted however. If OpenAI rarely actually uses such feedback then it would be a waste of my time of course. I realize that it is very hard to use though: 1) for feedback from arbitrary users openai won't even know the quality of their "improvements" over Codex, 2) I am working on a very complex project and the final 'diff' (that I nevertheless see as the correct response for Codex) would be very hard to verify manually without putting hours of (human) expert time into it. From the bullet point list that I had Codex generate, I think one already can conclude that my version is indeed better, but how to USE this data for subsequent training is even unclear to me (due to its size).