Question thread #147

[pauamma]

It's time for another question thread!

The rules:

- You may ask any dev-related question you have in a comment. (It doesn't even need to be about Dreamwidth, although if it involves a language/library/framework/database Dreamwidth doesn't use, you will probably get answers pointing that out and suggesting a better place to ask.)
- You may also answer any question, using the guidelines given in To Answer, Or Not To Answer and in this comment thread.

Comments

Re: This might be a bit long

[pauamma]

There are many aspects to your question, and I'm not able to address most of them, but I can make educated guesses for some of them and list the ones I'm not comfortable answering even tentatively:
1- I think the server-side API rewrite has stalled partway through, due to lack of senior developer, reviewer, and tester time. (momijizukamori, please correct me if I'm wrong about this.) Since, as far as I know, generative AI for code-writing uses is still at best at the level of a junior programmer whose code need to be carefully inspected and tested before being accepted, and the lack of reviewer (or senior developer wearing a reviewer hat) and tester is still a show-stopper for that approach. However, this may be a moot point, because...
2- Who has the copyright on AI output and therefore the right to place that code under GPL2+, as required to combine it with the Dreamwidth codebase is still, I believe, an unsettled legal issue in the US, to say nothing of the rest of the world. I think denise and mark will probably want to tread very cautiously here.
3- There are also ethical issues pushing against the use of generative AI for anything, which I will not address because I'm not qualified to.
4- The above, however, is distinct from developing API clients (the user-facing applications) using generative AI, using the API as it currently stands. I won't discuss this either, because I don't know whether or to what extent one could vibe-code a whole client or individual client features using the OpenAPI self-documentation feature of the Dreamwidth API as part of the prompt.

Re: This might be a bit long

[brickhousewench]

All things that we're discussing at my job as well. Oh well, I thought it was worth throwing out as a question, if only to get the conversation started on this question thread.

at the level of a junior programmer

I keep forgetting about this, as we don't have entry level programmers at my company. So all the AI coding at my job IS being supervised by senior developers.

Who has the copyright on AI output

Only humans can own copyright. I believe that the human who submits the code is presumed to be the owner, and thus can assign the open source license (My company is open source, and that's my understanding of how our legal folks are interpreting things, but YMMV).

ethical issues

I personally know people who are involved in some of the lawsuits about training LLMs on copyrighted works. And I have my own concerns about power and water usage with AI. But I pretty much have to use it if I want to keep my job, so I'm sort of gritting my teeth and getting stuff done.

Thanks for answering!