It would be one thing if this were only the case for professional development, or if AI were really being used as "just a tool" but it seems to be the dominant paradigm with AI everywhere, that LLMs have to "win" and all human involvement in coding has to be eliminated. A world where the dominant culture around LLMs was to actually use them for education and aiding code generation instead of entirely abrogating oneself to a black box dopamine machine would be infinitely better than the world we have.
At the core of Free Software there is one simple kernel: the user being in control. Licensing and copyright is used to achieve that, but they're the means, not the goal. The goal is that you, the user, should be in control of your computing, not the developers of the software. Hence software that is free i.e. the software not bound by the whims of its developers.
From that perspective, IMO, LLMs are far from antithetical to Free Software exactly because they let[0] their users do whatever they want with the software and their computers. That a user can instruct their agent to take the video playback code from MPV and combine it with the UI of XBMC[1] while ensuring it runs on the Raspberry Pi 4 they have connected to their TV even if they barely have any programming knowledge themselves is a sign of software freedom.
[0] at least the user-aligned ones not bound by restrictions placed on them by their own developers, so perhaps we need a Free Models movement too or something along these lines.
[1] random example, i don't even know of XBMC uses MPV's code already, so assume it doesn't
And because LLMs can't be audited and can't track or respect the licenses of the software they use as sources, no software generated by an LLM respects the freedoms of Free Software.
>That a user can instruct their agent (...) even if they barely have any programming knowledge themselves is a sign of software freedom.
It isn't, because as I said, freedom presupposes knowledge. If you barely have programming knowledge and you don't understand the code an LLM generates then you can't 1) read it, 2) edit it or 3) redistribute it without an LLM. You are not exercising any freedoms in that case, you're just being a consumer of a service.
>so perhaps we need a Free Models movement too or something along these lines
I said what I think would be the minimum requirements for a model to respect free software - at the very least it should only be trained on free software, it should be provably auditable without hallucination (I don't even think this is technically feasible) and it should respect the licenses of the software it incorporates, including attribution where necessary.
Beyond that there's still the problem that training even open weight models requires expensive infrastructure making it implicitly proprietary. As long as we need data centers at all it's never going to be as free as a text editor.
LLMs, even by the limits they have from their training, can be used to give more control to their users than what they have.
Yes, being able to train an open source (not just open weight) model that is fully aligned with the user would be the best scenario, but even without that, LLMs can still give users more control over their computing than they previously had - even within the confines of current copyright laws, if they so desire.
Because free software as a movement depends on the existence of licenses and copyright. Without the ability to legally enforce and defend those licenses it would be completely meaningless Just as any freedom is meaningless without a means to enforce and defend it. And as LLMs don't even allow the means, they can't be useful in reaching the goal.
The goal is for people to have control over their software - direct control, not control as determined by a mediator or service.
>LLMs, even by the limits they have from their training, can be used to give more control to their users than what they have.
We disagree on what "control" is in this context.
I don't believe prompting a service to generate code you can't understand and thus can't use without that service is control over code in any meaningful sense, any more than ordering food at MacDonalds is control over the menu.
>LLMs can still give users more control over their computing than they previously had - even within the confines of current copyright laws, if they so desire.
They can't, because LLMs don't respect those laws. It doesn't matter what license you use, it just gets sucked into the plagiarism machine like anything else.
Also, people had more control before LLMs because they didn't need to use LLMs. Knowledge is control. An empty text editor is infinitely more freeing than burning token on Claude (or whatever) will ever be.
Second, again, you're looking at things through the lens of licenses and missing the point that they're not the goal but the means of achieving control - and LLMs allow people to achieve that control in a different way.
Look, it is very simple: people who were unable to do that before (regardless of the reason) with an LLM can do take some piece of software and modify it to their own personal needs, or make some new piece of software or w/e they want. This is giving them more control over their computers than THEY previously had.
I.e:
Before: cannot make their OpenFoo do bar when they do baz by themselves, they have to cede that control over whoever works on OpenFoo.
Now: they can point OpenFoo's source to a (capable) LLM and instruct it (or have it guide them) to making it do bar when they do baz, all by themselves, without having to rely on whoever works on OpenFoo. The LLM can even provide advice on how to keep the program making bar when they do baz after OpenFoo updates.
That is IMO at the center of Free Software: people having the power of being in control of the software.
Sure, right now, there are some issues with the available LLMs and the most capable ones are controlled by others who in turn control you (e.g. Claude being prickly about security issues) but again open weight models exist and while the more capable of them require some very expensive high end hardware, not only capable models on (high end, for now) consumer hardware exists, but betting on the current status quo remaining as-is in the future is akin to thinking that "there is a world market for maybe five computers" (IBM chairman Thomas Watson in 1943) or that "there is no reason anyone would want a computer in their home" (DEC founder Ken Olson in 1977).
Ihe whole "Imaginary Property" (cute) charade is not rapidly headed to obsolescence. LLMs will only centralize control of IP to the corporations that will run their own LLM services, or license them to the services that will generate all of "your code" in the future. And under their terms of service, anything you generate will belong to those services, and not to you.
LLMs are antithetical to free software, which everyone seems to forget is as much an ethical and social movement as it is a classification of software licensing, and will be until one can examine and audit the training data of a model to exclude the possibility of including any non-free code, and until they include, document and respect the licenses for the sources they do use.
The degree to which they discourage people from actually being able to understand and modify code without an LLM also makes the code they generate implicitly nonfree. All of the freedoms of free software presuppose their utility to a human over their utility to a machine. If people aren't reading, writing and distributing code then it defeats the entire point.