Hi Hugo,
- Most of the of other arguments are so broad that they apply to any modern technology (search engines, the internet, gen-tech, money), probably even more so, and there I see no need to single out genAI.
I see one such reason: we can still opt out of genAI, whereas it's already too late for most of the other tech you mention.
All the tech you mention have turned into what Ivan Illich (in his book "Tools for conviviality") calls "radical monopolies" (not a term I like, but...). A radical monopoly by his definition is a technology that society at large has adopted to the point that individuals or small groups can no longer opt out of it without serious consequences. Whoever controls this radical monopoly thus sets the rules for everyone, and has the power to exploit everyone. You can go through your list and check for every example you mentioned.
If genAI becomes the undisputed norm in software development, it will reach the same status: it will no longer be reasonably possible to participate in a project like Guix *without* using genAI. We can watch this happen in other projects, which get drowned in huge vibe-coded pull requests. The choice they have is to reject such pull requests (and thus adopt some policy against some genAI use), or to embrace them and adapt their workflows, i.e. by using coding agents to review the pull requests as well. In the latter case, people without genAI will no longer be able to participate.
I see this GCD as a measure to keep Guix safe from creeping genAI obligation. It may be stricter than necessary for this purpose, nobody can tell. I am fine with revising the project's policy every year as we learn from others' experience. But history has shown that once you become dependent on some tech, it is very hard to get rid of that dependence again.
Cheers, Konrad.
