The UE Python libraries are a critical part of our strategy for building upcoming games. ChatGPT has done a great job authoring Python script for UE, but I often find that the approaches it uses makes use of deprecated APIs. This is especially frustrating because we’re already a bit restricted in our version lability due to PlayStation and Xbox compatibility requirements.
Is there an approach we should be using to stay up-to-date with Python without running afoul of other constraints?
Having to fix deprecation-related coding errors by hand is too time-consuming. Surely there must be a way to get ChatGPT working more reliably with unreal Python. Perhaps by creating more stable APIs that GPT can reference, for example?
Unfortunately there’s not much we can do to influence which API set ChatGPT would use when generating code. One suggestion would be to see if you can specify which API set it uses, as each version of the engine has a unique URL for documentation (e.g. Unreal Python API Documentation — Unreal Python 5.3 (Experimental) documentation). If that doesn’t help, there’s probably not much else you can do other than do manual post-processing on the generated code.
Thanks, @RajenK . Any chance of working with OpenAI to provide API guidance for how they can pin ChatGPT releases to a specific API version to deliver better retrieval-augmented generation?
If not, it would be helpful if you could point me to a PDF or other single-document source for the latest Python API (instead of a tree of web links, which ChatGPT currently rejects). It would save me from having to scrape your site.
