I think there is a deeper point regarding software freedom specifically - if a person is unable to understand the source code of something, then they are less able to modify it. It's why the definition of libre software centers around source code, even though modifying binaries is entirely possible. Despite being able to have a libre vs non-libre disassembler (analogous to proprietary vs open weight AI models), needing to use a tool at all diminishes your own ability to make changes to the code.

Now having said that, AI also increases software freedom in that it enables many people who aren't otherwise able to understand codebases (or even write simple code) to create/modify their own software. When we're talking about AI in the context of libre software, this is the constructive aspect we should be focusing on! Well defined plugin APIs that we know people are going to lay slop upon, etc.

But still, this doesn't have much bearing on a well-established fundamental systems project that needs to remain generally concise and human-understandable.