Just to name the takes that made me dislike today’s Linus:

https://lore.kernel.org/all/CAHk-%3Dwi4zC%2BZe8e%2Bp3tMv8TtG_80KzsZ1syL9anBtmEh5Z40vg@mail.gmail.com/

Quote:

It seems like [1] expresses a very anti-LLM position in general

Yes.

And no, that’s not the position of the Linux kernel.

I realize that some people really dislike AI, but this is an area where I’m willing to absolutely put my foot down as the top-level maintainer.

Linux is not one of those anti-AI projects, and if somebody has issues with that, they can do the open-source thing and fork it.

Or just walk away.

AI is a tool, just like other tools we use. And it’s clearly a useful one.

It may not have been that “clearly” even just a year ago, but it’s no longer in question today.

There’s a lot to unpack here, and it’s even worse than the shortened quote I had in mind when I wrote the above.

1. When he clearly had the good example of referring to LLMs in plain sight, he chose to refer to the slop machines as “AI” - especially in a context where it serves as a straw man. Most people opposed to LLMs do not have a problem with proper AI research, it’s just the statistical word prediction snake oil that we despise.
2. Using a straw man as someone who clearly has (at least for now) the intellectual capacity to understand that he is doing so is a really dirty way to argue.
3. For a project like the linux kernel, which has really become larger than just Linus, he threatens to override all opinions he doesn’t like on the topic (this is not the first time he’s done that). As we can see from the disappointing outcome of the debian vote on LLM contributions, this was a completely pointless hill to die on.
4. “Or just walk away” is children level of bickering
5. Again with the misnomer AI, to claim “[LLMs are] clearly a useful [tool]”, and to claim this “is no longer in question today” is a statement so dumb I have trouble believing he wrote that. I’ll give him that in some aspects LLMs can be useful, but the net effect on society (and that’s the larger picture he’s completely ignoring) may very well still be a net negative. This planet - even humanity - doesn’t need digital infrastructure to thrive. And 99% of the “usefulness” of LLMs is in maintaining or upgrading digital infrastructure and processing data. Something completely irrelevant to a thriving ecosystem. He’s clearly locked into “tech bro think” where it doesn’t even occur to him to consider impacts on the environment and weigh them against the perceived usefulness of LLMs. Not to mention that you do not know whether an LLM will be useful in a particular case until after you have solved the case - at which point you may say “oh great, it helped me” or “oh shit, the LLM wasted a lot of my time on this”. And if you can’t tell in advance whether for a particular scenario a given tool will be useful or not, then using it will also waste people’s time and it’s absolutely fair to expect that people - even when isolating their views to the tech sector - weigh the perceived usefulness against the time wasted and evaluate whether there’s actually a net positive.

I am pretty sure - also from personal experience - that the next benefit of using LLMs for software development only marginally outweighs the downsides (wasted developer & computing time, wasted resources). And taking the social, political and environmental impacts into account, the currently hyped machine learning applications are a HUGE net negative for mankind.

Plus which, LLMs have already peaked, because they are all out of unpoisoned training data. So things will get worse from here on out.

Accordingly, my verdict is that Linus is losing his marbles. Because he once was smart enough to realise all those implications. And now he’s decided to gloss over them and join the hype train.