You can read it straight from the person who coined the idea.
In short: these are computer programs whose authors allow you to use, modify, copy, and redistribute the modifications freely. For this to be possible, the program's source code must be available to you. In other words, they are programs where the user has full control.
Following the core principles that inspire free software, it is a computing model applicable to programs, protocols, formats and standards, services, and user experiences in general. It follows a set of fundamental principles such as independence, privacy, and the freedoms to use, copy, modify, and redistribute. Free computing goes beyond software — it also affects standards and relevant data sets or models.
By using it.
Besides: explaining what it is, how to use it, sharing your experience… writing about or spreading that knowledge in a respectful way. Also by developing free software yourself, or contributing code, documentation, etc., to existing projects.
Free software has no political color. Although it has often been associated with the left, many aspects of free software could be attributed to a liberal or right-leaning worldview. In reality it is a politically neutral system, based on game theory, the free market, individual freedom… and with an implicit foundation of social solidarity. It defies political classification, as shown by the fact that politicians across the spectrum have promoted it in very different social contexts.
Both terms refer to the same programs. But I prefer to use the words that best describe what I mean — that way we avoid confusion.
Well… first of all, I am not an activist. I’m much more interested in the path of leading by example. More on that topic here.
Defending causes doesn’t make much sense. Discovering their true relationship with their actual effects, perhaps. But that’s a personal task.
Ad populum fallacy. And besides, I need to be convinced I’m wrong before I change my mind.
These days people say “open source AI” to refer to freely distributable LLMs, which is something very, very different from “open source”. It’s useful to call things by their proper name so we don’t lose our bearings.
It’s symbolic of your struggle against reality.
(…and you’ll lose “your fight” anyway)
Yes. For the sake of humanity’s own progress, software should not be patentable.
If you’re referring to LLMs, agents, and derived projects/services, I’m very interested in the scientific side. As products, I only care about LLMs that are used with free software — with an important caveat: even when some of these models are distributed for free, we don’t know the data sources they were trained on. That’s why it’s wrong to call them “open source,” and even more so “free.” Unless you build those models yourself (possible but impractical for large ones), there is still no general-purpose LLM that qualifies as truly free (“a free AI”) that has been published.
