>> What is (or should be) the policy for AI/LLM usage as it relates to content under scheme.org? I've been thinking it could be productive to let a
>> robot create entries for index.scheme.org (of course, it would still be reviewed by human eyes). But I'm not going to do this without approval
>> from others.
>
> I realize that my opinion is not really meaningful, as I’m not an
> official person in relation to scheme.org, so feel free to ignore.
>
> However, as a scheme-index contributor, I’d feel very disinclined to use
> and contribute to it if I know that the data was contaminated by an
> LLM. The reasons have been discussed elsewhere, from copyright
> infringement risks to security malpractice to environmental impact.
>
> As an alternative to using such a technology, I suggest you first open
> respective issues on scheme-index so that an accountable human
> contributor (like myself) has an opportunity to nerd snipe themselves to
> it. I’d be very much happy to help if I know that it keeps the project
> clean and humane.

I do not want to influence the daily maintenance of scheme.org, but I initially required certain rules, and as Arthur mentions, there is a charter. This charter was, however, written before LLMs were widely used, and even though I'm a computational linguist myself and have worked with both machine learning and linguistics, language models were very specialised tools until recently, and it was difficult to predict the current causal and widespread use of LLMs. If the charter would have been written today, I would certainly have made sure it included some language about LLMs. Since I have not thought about the specific intersection between scheme.org and LLMs, I cannot produce a policy text worthy of inclusion right now even if I wanted to, but I can say a few things. Please excuse me if my thoughts are not entirely coherent.

My feelings are fairly aligned with what Artyom writes here. I think that one document that everyone working with programming and programming languages should read, even if they don't agree with it, is the Codeberg blog post "Protecting our FLOSS commons from LLMs" (https://blog.codeberg.org/protecting-our-floss-commons-from-llms.html).

The most important factor in any community is the people who want to contribute to it. I understand that some people would want to contribute using LLMs, and may not even be able to contribute without, or at least feel that they can't. Then there are those who would contribute less or nothing at all if others are using LLMs. An LLM policy would have to address this.

Also, there is a significant toll on society from LLMs, not least from an artistic, economic and environmental perspective. I think that cannot be ignored. On the other hand, there are people who genuinely benefit from using LLMs. I do think most people overestimate how useful they are, which is natural if you don't know the mechanisms involved, and I believe most people would be much better off with other types of tools. That said, I'm not the one that runs this community, so I'm not equipped to understand all nuances, requirements and feelings in it.

Hope this helps,
Magnus