AI/LLM and its place in a OSS project #1268
kennethshsu
started this conversation in
General
Replies: 0 comments
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
With the rise of more and more powerful AI/LLM models, I've been thinking a lot about bad practices and what keeps a community project "good" and healthy.
And, this is what I came up with:
Your AI generated output should only be consumed by you, not others.
This applies to everything, code, questions and answers, revision, review feedback, ideas, drawings, documentation, whatever. If anything is generated by AI under your direction, YOU consume the result, don't pass it without digesting it first. And it's only after you digest the result, then you are adding value.
There are some tickets and pull requests that are completely written by AI, they are SOOO expensive (in reviewers' time) to read and consume. We also have no idea on the prompts used. For example, was it "find five missing features in this package and open five PRs on those features", or "tell me which lines to look at that can point to a bug in the parallelogram calculation"? I think it's great to use AI and coding agents, but if you write the prompt, you need to consume the result yourself. One doesn't just pass that result for others to consume.
If the above becomes a rule/guideline, I think it will help a lot with "AI slops" and also having to just interact with AI. It will also turn this community into more of a learning one, which is what I'd hope for. I also absolutely hate to see in some instances (on other projects) that the code is written by AI, PR is written by AI, reviewer uses AI, and responses are also AI. IMO this is not a "community" but just bots dancing together stupidly. The only value add to the project is literally just the amount of tokens spend (i.e. pennies). I also don't understand the motivations from these users that heavily rely on AI, what are they hopping to get out of? That they spent a few dollars and use a cool AI model?
AI is a tool, but only a tool, it does not replace human.
Anyways, this is just an idea/philosophy at this time. But I'm curious what other people think. Let me know if you agree or disagree 😃
All reactions