Experimenting with Chat-GPT4 for generating code

[julianharty]

Context

I gather many people are already using "AI" to generate source code and there are numerous articles written about the effects of using Chat-GPT and/or GitHub co-pilot. On a smaller scale I'm interested to learn more about using Chat-GPT4 in particular in terms of any increase in productivity and in terms of the quality of the code and the results.

Objectives

• To be explicit in identifying the use of Chat-GPT4 including providing feedback to it in order to discover whether it can improve flawed code.
• I'm unlikely to ask it to integrate its proposals into a larger codebase.

Considerations

The following considerations are instances of what's likely to be a larger scale set of considerations for people who use Chat-GPT4 to create code. They're recorded here to indicate they're being actively considered.

1. This repo is freely available as opensource, is code that was generated by Chat-GPT4 acceptable as opensource software from a licensing perspective? And who owns it?
2. What are the effects of that code being modified by humans in terms of copyright and licensing?
3. What about the effects of Chat-GPT4 learning from many opensource projects that may have their own licensing constraints?