Hi Ludo' and all,
Thanks again for taking up this important GCD.
Leaving the pledge itself aside for a bit, I like to focus on the
arguments leading up to it (since the pledge depends on those). To me,
the arguments fall into two categories:
- Most arguments I full agree with, e.g. the "model building" is
essential, and vibe coding without even reading is bad, etc. There is
plenty of possible genAI use left that is not argued against.
- 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'm mostly susceptible to the "using genAI is supporting
facism"-argument. I believe it to be hyperbole, but am willing to be
convinced otherwise, so let's assume it's true.
Most other modern technology is a tool for facism as well. So if we go
the route of pledging against using "fascist tools" (and I can be
convinced), we should do it properly, and pledge against the bigger ones
first.
At least, I cannot with a good conscience join the pledge as written,
because it misses the forest for the trees.
vvvvvv Start Of Rant About My Favorite Fascism Tool vvvvvv
The most important fascist-tool is in my opinion our debt-based and
interest-bearing monetary system. Almost all of our money comes into
existence through loans. That is, for every euro we have and spend,
there is an associated loan that someone has to pay interest on. And
the interest-paying party is generally someone poorer than the one
spending it, funneling money from poor to rich, world-wide.
I'm willing to argue that our current monetary system is the leading
cause of most of our other problems (wars, climate change, polarization,
etc.), to the extend that most of the problems we see with genAI would
disappear if we fix the money first. (This is not the place for that
discussion though.)
If we are going to pledge against using tools that can be used for
fascism, then I urge us to start with the elephant in the room and
pledge to not use debt-based money. That is, use only resources freely
given to us, traded for in kind, or paid for through alternative currencies.
^^^^^^ End Of Rant About My Favorite Fascism Tool ^^^^^^
To be constructive, I propose to do either of the following (or both):
- Allow genAI use that fosters our human connection and our theory
building etc. For examples see "Where AI Actually Helps" in the cited
article about theory building.
- And/or, amend the pledge to not use any tools that are used for
fascism. In particular, we should pledge to not use debt-based money
for anything Guix-related (and we should pledge not to fly).
The rest of this message is me addressing each argument individually.
The primary reason for that is to have each issue explicitly on the
record, so it is clear at the deliberation period that I have
articulated all my concerns during the discussion period.
I'm willing to discuss any of the points below in more detail, but I
think it is more important to discuss the big picture first.
Hugo
On 5/15/26 17:19, Ludovic Courtès wrote:
However impressive the results may look, the authors believe genAI has a
social and environmental impact that undermines the very humanist
foundations of free software and Guix:
I'd argue the opposite, that, if used well, genAI empowers people to
join open source projects, thereby enabling humanism. For example genAI
is a powerful tool to learn and explore and discover and is especially
well suited to open source software projects.
Another more basic example, we reach about 5 billion people with the
current languages we support. (Which is quite impressive actually.) If
those other 3 billion people can use genAI to communicate with us more
easily, then that is a big win. Yes it would be better if we have
translators for all languages, but we don't.
It's not all roses of course, I can see that too, I'm just not willing
to throw away the baby with the bathwater at this point in time.
- The leading commercial genAI services run non-free software,
Yeah the models and services being closed source is bad; we should
resolve that, and I believe we will as society.
I also believe that should not matter w.r.t. to the artifacts we
produce. We probably have many contributors that use non-free software
to contribute to Guix.
are a threat to user privacy,
I don't think user privacy is really applicable here, because everything
we do w.r.t. Guix is already public. I probably have shown more of
myself through these discussions than through any other medium. Also,
contributors can decide to not use genAI if they fear it affects their
privacy.
and achieve a concentration of power
Well in general maybe, but not for guix. We should be good as long as
we do not become dependent on genAI, and I believe that should be doable.
over the lives of people rarely seen before.
This is a bit hyperbole, we are talking about open source software
development. Besides, other technology like search engines and money,
are worse, so let's pledge against those first.
- GenAI launders the reciprocity baked into copyleft licenses such as
the GNU General Public License (GPL), effectively violating it. A
real-world example of copyleft-laundering is [the `chardet`
LLM-assisted “rewrite” for the stated purpose of relicensing from
LGPL to MIT/Expat in March
2026](https://tuananh.net/2026/03/05/relicensing-with-ai-assisted-rewrite/)
([covered by LWN](https://lwn.net/Articles/1061534/)) or the [EmDash
WordPress reimplentation in TypeScript “under the more permissive
MIT license”](https://blog.cloudflare.com/emdash-wordpress/).
Yes, so let's not do copyright laundering. I don't think anyone was
planning to.
- The narrative around the output of genAI tends to demean human
creativity, to [discourage the learning and knowledge sharing
processes](https://cekrem.github.io/posts/programming-as-theory-building-part-ii)
at the core of free software,
Yes, let's keep that 'theory building' part alive. Please note that
that page has an entire section called "Where AI Actually Helps (With
Clear Constraints)". That section quite succinctly describes how I
think we should use genAI.
and to [discourage free software
contributions by humans](https://arxiv.org/pdf/2601.15494).
Yes, we should ensure genAI use fosters "engagement", not weaken it.
(Disclaimer: I only read the abstract.)
We
believe it [stifles individual
autonomy](https://ali-alkhatib.com/blog/defining-ai)
That use of genAI is so far detached from how we could employ genAI that
this article seems totally meaningless in this discussion. How I
anticipate we can use genAI, would not even pass their definition of AI.
This argument, or at least its justification, is therefore not
relevant as far as I can see.
at a
fundamental level—replacing one’s ability to build up knowledge with
a false sense of quick achievement,
I suppose it could happen that people stop learning because they have
genAI. Certainly something to stay aware of. But it is also possible
to use AI to learn. I use it to explore ideas and expand my horizons.
I also learn through other mechanisms, but genAI is good too.
Yes I also use genAI occasionally for a 'quick achievement', but I also
occasionally eat fast food, etc. This argument is way too parentalistic
to be included.
building up [cognitive
debt](https://simonwillison.net/2026/Feb/15/cognitive-debt/)—while
Yes, we should use genAI to *reduce* debt. This article describes vibe
coding without even reading the code. Yes we should pledge to fully
understand all the code we incorporate in Guix.
also [weakening communities and destroying
labor power](https://tante.cc/2026/04/21/ai-as-a-fascist-artifact/).
Thanks for including this. This article has arguments that I'm
receptive to. Nevertheless, the same arguments apply to most
technology, while at the same time they apply very little to a project
like Guix. See above.
- The huge ecological footprint of genAI is well documented, going
from the
[energy](https://cleanview.co/content/power-strategies-report),
[water, and materials needed to build the servers used to train
models and to service
requests](https://www.technologyreview.com/2025/05/20/1116327/ai-energy-usage-
climate-footprint-big-tech/),
to the land usage of those data center, and to the energy and [water
needed to operate those data
centers](https://theconversation.com/data-centers-consume-massive-amounts-of-
water-companies-rarely-tell-the-public-exactly-how-much-262901).
Fair sharing of natural resources and the habitability of the planet
are prime concerns from a humanist perspective.
Well yes, I fully agree with the last sentence, but the part before does
not follow.
I calculated the energy and water use of my Guix contributions, and the
majority of it is (re)building packages and my physical existence. GenAI
is a net positive, because it reduces the number of rebuilds I have to
do. (In my limited experiments, codex fixed packages in less attempts
than it would take me.)
And more importantly: the energy use is just not that much compared to
other tech like flying. We should have a pledge that none of us will
fly 'for Guix', and ask people why they felt the need to fly if they did
so anyway. And use that to learn how we can organize such that people
don't want to fly. This suggestion is not even sarcasm, we should limit
flying, it is one of the most important (and easiest) things to do to
fight climate change. Once we limit flying, we can look at other
factors, otherwise it is "dweilen met de kraan open".
Luckily we already have vegan food at the Guix days. Thanks for that.
To this humanist perspective, we should add legal concerns:
I might address the legal concerns later, but here I think we should
probably do what experts advice us.
Hugo
