Hi Ludo and All, Thank you for GCD-008. I appreciate the work that went into this GCD and I share some of the concerns raised here. However, regretfully, I feel I need to express my general dissent with it. I think some of the GCD's premises are unclear or weak and I do not agree with the proposed goals.
Below, I indicate the points that I find most problematic in the current version of the GCD. In general, I think most of the arguments presented in the GCD equally apply to Generative AI as well as other technologies that are regularly packaged or otherwise used by Guix. Of the arguments that might be peculiar to Generative AI, I think they might justify restricting some specific uses of Generative AI but not a complete ban. > The leading commercial genAI services run non-free software, are a threat to > user privacy, and achieve a concentration of power over the lives of people > rarely seen before. Does "running non-free software" here refer to training models or to offering Generative AI services? A similar criticism applies to other technologies that are regularly packaged in Guix. For instance, web browsers routinely interact with proprietary services, raise significant privacy concerns, and concentrate power in a small number of organisations. Furthermore, would this argument be a reason to ban Generative AI-based contributions in Guix, to avoid packaging Generative AI tools, or to avoid packaging projects that have been developed using Generative AI? What about using LLMs that do not come from leading commercial services? > The narrative around the output of genAI tends to demean human creativity, to > discourage the learning and knowledge sharing processes at the core of free > software, and to discourage free software contributions by humans. Does this argument refer to the narrative or the technology itself? If it's the former, then the narrative is separate from the technology - it is possible to use the technology without buying into the narrative. > The huge ecological footprint of genAI is well documented, ... Guix does not, as a rule, refuse to package energy-intensive software. Ecological footprint concerns may apply to non-Generative AI tools and libraries that are packaged in Guix, such as some Data Science or electronic currency applications. Furthermore, the ecological argument would not apply equally to small and large LLMs. In my opinion, it would be difficult to use this argument to single out Generative AI and apply a blanket ban to it. > At the time of writing, only proposed interpretations of copyright law exist: > that depending on the level of human intervention, genAI output could be > considered not copyrightable... The legal concern would be a basis to (possibly temporarily) reject Generative AI contributions, but I am not sure it could justify refusing to package Generative AI tools and projects developed using Generative AI. Separately, while I see potential Intellectual Property concerns when it comes to core parts of Guix, I wonder how significant the IP risk would be for more formulaic contributions such as (simple) package definitions and package updates. These are the main points on which I am not aligned, or which I find unclear or ambiguous in their current wording. I regret having to play the contrarian here, but I hope this might be a useful contribution to the discussion. Best wishes, Fabio (fnat). -- Fabio Natali https://fabionatali.com
