GNOME's default Image Viewer (Loupe) no longer accepts AI generated contributions. You can read more about the reasons for my decisions in my discourse post.
I'm planning to adopt the same policy for the rest of the projects I maintain. And I think others should as well.
https://discourse.gnome.org/t/loupe-no-longer-allows-generative-ai-contributions/27327
@sophie It's interesting idea, how do you plan to enforce and distinguish between AI and non-AI?
I would be very afraid of discrimination based on slander that the person uses AI to contribute, which effectively (except more and more rare cases of old or low-power LLM) cannot be proven.
@sophie, I don't deny that LLMs (large language models) consume a huge amount of resources, but computers themselves do too. Can't we just print the picture and look at it instead of burning power to present it with Loupe?
It seems slightly unreasonable to "cripple" the GNOME ecosystem when Apple and Microsoft are allowed to leverage AI.
What about educating contributors on how to use AI more efficiently to contribute, build local datasets based on FOSS-licensed software, etc.?
@kolunmi @sophie it's significantly worse now. It's in infancy. Same as room computers back then.
I address the moral part with developing LLM trained on OSS licensed data (I would say that's good way how to share knowledge).
I have used Windows, it's terrible and annoying.
Why majority of users using it then? Microsoft seems to still offer something GNOME doesn't.
@lhp @sophie as a FOSS developer, I have to tackle tasks outside of my expertise, so yes. Sometimes I don't understand the task. I can read many blogs, learn from multiple sources, or just verify (basic logic and references) answer given by LLM or LLM search. Depends how much time I can spend on the task.