You’re getting a lot of flack in these comments, but you are absolutely right. All the concerns people have raised about “AI” and the recent wave of machine learning tech are (mostly) valid, but that doesn’t mean AI isn’t incredibly effective in certain use cases. Rather than hating on the technology or ignoring it, the FOSS community should try to find ways of implementing AI that mitigate the problems, while continuing to educate users about the limitations of LLMs, etc.

Luddites were not as opposed to new technology as you say it here. They were mainly concerned about what technology would do to whom.

A helpful history right here: https://www.hachettebookgroup.com/titles/brian-merchant/blood-in-the-machine/9780316487740/?lens=little-brown

I dont think the community is generally against AI, there’s plenty of FOSS projects. They just don’t like cashgrabs, enshittification and sending personal data to someone else’s computer.

Is there no electron wrapper around ChatGPT yet? Jeez we better hurry, imagine having to use your browser like… For pretty much everything else.

Good.

The Luddites were right.

Gnome and other desktops need to start working on integrating FOSS

In addition to everything everyone else has already said, why does this have anything to do with desktop environments at all? Remember, most open-source software comes from one or two individual programmers scratching a personal itch—not all of it is part of your DE, nor should it be. If someone writes an open-source LLM-driven program that does something useful to a significant segment of the Linux community, it will get packaged by at least some distros, accrete various front-ends in different toolkits, and so on.

However, I don’t think that day is coming soon. Most of the things “Apple Intelligence” seems to be intended to fuel are either useless or downright offputting to me, and I doubt I’m the only one—for instance, I don’t talk to my computer unless I’m cussing it out, and I’d rather it not understand that. My guess is that the first desktop-directed offering we see in Linux is going to be an image generator frontend, which I don’t need but can see use cases for even if usage of the generated images is restricted (see below).

Anyway, if this is your particular itch, you can scratch it—by paying someone to write the code for you (or starting a crowdfunding campaign for same), if you don’t know how to do it yourself. If this isn’t worth money or time to you, why should it be to anyone else? Linux isn’t in competition with the proprietary OSs in the way you seem to think.

As for why LLMs are so heavily disliked in the open-source community? There are three reasons:

1. The fact that they give inaccurate responses, which can be hilarious, dangerous, or tedious depending on the question asked, but a lot of nontechnical people, including management at companies trying to incorporate “AI” into their products, don’t realize the answers can be dangerously innacurate.
2. Disputes over the legality and morality of using scraped data in training sets.
3. Disputes over who owns the copyright of LLM-generated code (and other materials, but especiallly code).

Item 1 can theoretically be solved by bigger and better AI models, but 2 and 3 can’t be. They have to be decided by the courts, and at an international level, too. We might even be talking treaty negotiations. I’d be surprised if that takes less than ten years. In the meanwhile, for instance, it’s very, very dangerous for any open-source project to accept a code patch written with the aid of an LLM—depending on the conclusion the courts come to, it might have to be torn out down the line, along with everything built on top of it. The inability to use LLM output for open source or commercial purposes without taking a big legal risk kneecaps the value of the applications. Unlike Apple or Microsoft, the Linux community can’t bribe enough judges to make the problems disappear.

Reminder that we don’t even have AI yet, just learning machine models, which are not the same thing despite wide misuse of the term AI.

You can’t do machine learning without tons of data and processing power.

Commercial “AI” has been built on fucking over everything that moves, on both counts. They suck power at alarming rates, especially given the state of the climate, and they blatantly ignore copyright and privacy.

FOSS tends to be based on a philosophy that’s strongly opposed to at least some of these methods. To start with, FOSS is build around respecting copyright and Microsoft is currently stealing GitHub code, anonymizing it, and offering it under their Copilot product, while explicitly promising companies who buy Copilot that they will insulate them from any legal downfall.

So yeah, some people in the “Linux space” are a bit annoyed about these things, to put it mildly.

Edit: but, to address your concerns, there’s nothing to be gained by rushing head-first into new technology. FOSS stands to gain nothing from early adoption. FOSS is a cultural movement not a commercial entity. When and if the technology will be practical and widely available it will be incorporated into FOSS. If it won’t be practical or will be proprietary, it won’t. There’s nothing personal about that.

There are already a lot of open models and tools out there. I totally disagree that Linux distros or DEs should be looking to bake in AI features. People can run an LLM on their computer just like they run any other application.

Sounds like something an AI would post. Quick, what color are your eyes?

I’ve yet to see a need for “AI integration ✨” in to the desktop experience. Copilot, LLM chat bots, TTS, OCR, and translation using machine learning are all interesting but I don’t think OS integration is beneficial.

…this looks like it was written by a supervisor who has no idea what AI actually is, but desperately wants it shoehorned into the next project because it’s the latest buzzword.

I won’t rehash the arguments around “AI” that others are best placed to make.

My main issue is AI as a term is basically a marketing one to convince people that these tools do something they don’t and its causing real harm. Its redirecting resources and attention onto a very narrow subset of tools replacing other less intensive tools. There are significant impacts to these tools (during an existential crisis around our use and consumption of energy). There are some really good targeted uses of machine learning techniques but they are being drowned out by a hype train that is determined to make the general public think that we have or are near Data from Star Trek.

Addtionally, as others have said the current state of “AI” has a very anti FOSS ethos. With big firms using and misusing their monopolies to steal, borrow and coopt data that isn’t theirs to build something that contains that’s data but is their copyright. Some of this data is intensely personal and sensitive and the original intent behind the sharing is not for training a model which may in certain circumstances spit out that data verbatim.

Lastly, since you use the term Luddite. Its worth actually engaging with what that movement was about. Whilst its pitched now as generic anti-technology backlash in fact it was a movement of people who saw what the priorities and choices in the new technology meant for them: the people that didn’t own the technology and would get worse living and work conditions as a result. As it turned out they were almost exactly correct in thier predictions. They are indeed worth thinking about as allegory for the moment we find ourselves in. How do ordinary people want this technology to change our lives? Who do we want to control it? Given its implications for our climate needs can we afford to use it now, if so for what purposes?

Personally, I can’t wait for the hype train to pop (or maybe depart?) so we can get back to rational discussions about the best uses of machine learning (and computing in general) for the betterment of all rather than the enrichment of a few.