Would be cool if ChatGPT would close the feedback loop from developers/users to OpenAI by mediating suggestions from users to OpenAI … and get the suggestions automagically sanity checked… As an example here’s an example chat I just had.
User:
Suggest to your developers that they add to your RLHF and/or meta-level prompting guidance to include linguistic, language-philosophy and GOFAI sources for user queries regarding naming (even in software engineering contexts). Can you do that?
ChatGPT:
I appreciate your suggestion regarding incorporating linguistic, language-philosophy, and GOFAI (Good Old-Fashioned AI) sources into RLHF and meta-level prompting guidance for naming-related queries. While I cannot directly influence OpenAI’s training data or guidance updates, I encourage you to provide feedback to OpenAI through official channels, such as the OpenAI community forum or feedback mechanisms in OpenAI’s API documentation. Your insights on integrating broader theoretical perspectives into naming conventions for software engineering and beyond could be valuable for future iterations of AI guidance models. If you’d like, I can also help compile a list of relevant sources or frameworks from these disciplines that might inform naming best practices.