I believe that the proAI/antiAI from the point of view of users and developers are widely discussed.
But I want to consider the problem from a strategical point of view.
There are basically three potential positions:
- Accepting LLMs as tools (with safeguards or not).
- Choosing not to do anything for now except some minor guidelines (formally Debian and the current position of DC), aka the “middle ground”
- Rejecting LLMs strongly (with potential exceptions that have major safeguards)
What we could learn from Debian is that, basically, nobody is really happy with that. Also, it looks like it is not a sustainable position. Projects telling 2) will gradually slips towards 1) through contributions made by LLMs. The longer they wait, the more ineluctable 1) is.
Which means that a strong choice is probably required. 1 or 3.
Both choices will alienate some part of the community and some key contributor. That’s sad but a small niche project cannot please everyone.
Now that I made this point, we way argue whether 1 or 3 is better from a philosophical, ethical or pragmatic point of view.
But, from a strategical point of view, what will be the effects on the project:
PRO-AI
PRO. contributors can use LLMs which may be more productive in the short term (while every studies so far tends to point the opposite)
CON: alienate part of the community with is mostly anti-AI and look for ethical software
ANTIAI:
PRO:
- strong support from the “ethical” part of the community
- potential better long term maintenance (not depending on proprietary stuff)
CONS:
- Alienate some contributors that really want to use AI
When you think about it, contributors that are completely turned away by the fact they can’t use LLMs are probably contributors you don’t want in your project, even if you have position 2) “moderate use of AI” like Debian. People that don’t want to contribute without LLMs would not be able to humanly review and understand the code anyway.
Also, Delta Chat is a very niche project currently appealing to open source/decentralized/privacy minded people. Those people usually have a very strong ethical sense. They are very vocal about supporting “projects not using any AI”. They may be a very small minority of the world but in the DC niche, they are very influent. By appealing to those people, Delta Chat, with all its other qualities, could become a major player and the one those people highly recommend (for example, Simplex was at some point quite popular with strong privacy minded people but is now replaced due to suspicion of its creators being fascists).
In conclusion, I would say that Delta Chat has all to gain from adopting an anti-AI stance and nothing to really lose. Cherry on cake: being anti-AI is a policy that can be changed later if we realize the world evolves toward ethical and reliable LLMs (try to read this sentence again but without laughing).
While the opposite is false: once you have have slop in your codebase, it will be very hard to remove. Cory Doctorow compares it to asbestos.
EDIT: did a cleaner version on my blog The AI Pascal’s Wager