Towards an AI usage policy - your thoughts please

submitted by
he/him

For too long PieFed has been without a policy on using LLM/AI-generated code in PieFed.

My attitude remains basically the same as that I expressed recently - https://piefed.social/comment/10688199.

I am very close to publishing a policy so any interested parties - this is your opportunity to speak up. Please send me a private message if you do not feel comfortable putting yourself out there - this can be a contentious topic and if you don’t want to deal with people getting all up in your grill about it, that’s totally understandable.

If you’d like to do more reading and thinking about this, here are some links that I found helpful lately:

9
17

Log in to comment

Hot Top New Old
edited

I am in the camp of no AI contributions due to the output not confirming to the projects licence.

Would you accept a contribution under BSL or SSPL?

 reply
7
he/him OP

This is in a medical setting but has obvious parallels to coding: https://benngooch.substack.com/p/i-was-an-enthusiastic-early-adopter

 reply
3
edited

Just a blog i read : https://github.com/Vxrpenter/AIMania/blob/main/WHY.md

 reply
3
he/him

That is incredible, I think, I will be using this more often, because arguing again and again about the same few points is exhausting.

 reply
2

Glad it can help :)

I was lucky, in the collective kittens we love sharing and reading those links :)

 reply
1
[any] edited

I thought about it for a while. And I think my main point wouldn’t even be on a factual level. I mean sure, AI is bad and all. For many different reasons. But most importantly, PieFed is a free and open-source project and literally nobody here likes AI contributions. A good chunk of the userbase dislikes it. Seems the developers all(?) dislike it. And there isn’t anyone around willing to review and merge these things. So I don’t see any reason to argue here. Just don’t send in PRs with stuff everyone dislikes.
And on a second note, also please don’t try to deceive volunteers either, and sneak it in or something… That should be basic respect.

That’s more or less the “why”. The “how” is probably way more complicated. People want exact rules. But the lines are blurry. AI can be used to just assist. Like explain an SQL query, or come up with one. Maybe using it once in a while isn’t worse for the environment than some silly CI pipeline… And a query, or database migration script won’t really touch copyright either. Maybe someone used Claude for some red teaming and the exploit actually works… These exact contribution guidelines tend to get quite detailed.

 reply
2

The policy of requiring AI code to be marked seems reasonable to my non-coder eyes.

 reply
2

PieFed is my new home and I’ve been loving it. Please don’t allow AI code. I really don’t want to leave the platform, but that would be a deal breaker. I cannot stress enough, fuck AI.

 reply
3
he/him edited

As I have previously expressed, I am firmly against allowing AI contributions to our project mainly due to ethics reasons.

In my opinion, we as a quite political project, that tries to radically rethink social media as a self-governed, community-led project out of fascists control, can’t simply put our head in the sand and ignore what is happening around us.

At the end of the day, AI is mainly a tool for capitalists to take all of our hard work, strip it of all of it’s intent and meaning and reproduce it without paying us a dime or even acknowledge where it came from. We as free software devs should value and appreciate each others hard work, not just exploit it without permission. There is already projects to “free” GPL licenced code, by just reproducing it with AI to not have to abide by it’s terms.

Additionally, AI significantly harms the people (that are hired to filter and fine tune AIs for inhumane pay) and the environment.

Producing code with AI makes us understand our code less, there is a huge difference between actively writing code and checking code, someone (or something) else produced. This will inevitably lead to worse, more insecure code, even if experienced devs use it.

AI can also make developers dependent on it, as we loose the understanding of the nitty-gritty details over time, because we don’t have to think about them anymore. This is actually exactly, what the companies are designing it for. We should foster a strong developer community, that is able to properly maintain this codebase independently from any AI. Because AI can’t stay this cheap, and if we depend on it to much, that might fuck us over.

Due to all of these reasons and more, I don’t think of AI as a neutral “tool” as Linus Torvalds sees it. A tool isn’t necessarily neutral, a nuclear bomb is also just a tool. I think it is increasingly important morally, but also practically for our long term survival as a project, to explicitly reject the use of this tooling for producing code or participating in discussion.

This is also not some wild, out there take, quite a few projects have taken this step and it seems to have worked out pretty well for them. An example, I could think of is servo (https://book.servo.org/contributing/getting-started).

I know, this would be hard to enforce, but it’s also about taking a stand and maybe making some people think twice about making a lazy slop contribution. This shouldn’t come at the expense of people making real contributions of course. If were not really sure, we should give people the benefit of the doubt, because having to prove that you really made the code would also be really annoying. But we could still stop the worst, most obvious cases. And it would be a step in the right direction, I think.

 reply
1
Insert image