Linux

as a pro-AI, anti-cloud nerd? All of these problems disappear when we address the real problem of Capitalism.

there are abandoned tech certs for building datacenters that produce 110% of their own electricity. these are distinct from the data mines these useless nerds want to build.

without capital, the surveillance state dies.

intellectual property is a capitalist myth invented to hold us back and induce rents, and I'm tired of pretending otherwise.

oh no, you fired me from a job I'd rather not have to do and I still get to survive because we slew capitalism? ahhh, how will I cope with the blow to my identity?

It's that "greater precision" part that is hard to ensure.

If someone is checking their work with AI or thoroughly checking the work after, great.

But it's a hard difference to teach

The comparison between AI and compilers is kind of stupid though. When you program in a given language, you're writing out the entire logic and behavior of your program. Not to mention that the compiler is a deterministic machine, it will always behave in a certain manner and produce predictable results. On the other hand, using AI, you just write a sentence. Then it "creates" a whole bunch of code that will not always be the same, even if you always use the same model and prompt. Not to mention all of the detrimental effects AI has, cognitively speaking, when you start to rely on it. On top of this, all AI models were trained on unfathomable amounts of stolen information.

His line seems to be ‘don’t submit shit code.’

And he's been 100% consistent with that since the start.

Accept something along the lines of a positive account given by an experienced software engineer - it’s useful, it presents risks but also it presents new ways to get certain work done well in much less time etc.

I agree to the utility of finding security bugs but the general utility of using LLMs is often overblown. There are studies that say that 95% of companies do not see any measurable positive impact.

The Linux kernel is likely a special case. But for other open source developer communities, it seems to have some negative impacts, too. And the full effects on maintainability of code will only be visible after some years.

I dislike LLMs/GenAI mostly due to ethical reasons (I'm aware the code quality from LLMs has increased over time), but I've already accepted the fact it's gonna be used a lot. In most cases, I can avoid software using it in favor of alternatives, but that's not always possible, such as Linux itself. I'm not gonna risk staying on outdated kernels, and I don't have the time and knowledge to maintain a fork of a whole kernel.

As for my own code, I refuse to let AI touch it because I like thinking about how to code something, and I get really happy when it works.

This mindset is the entire problem. "Oh we don't need to worry about ethics or morals, just efficiency" is exactly the issue with every pro-AI argument and every pro-capatalism argument. Its the issue with our governments, with investing, with corporations.

I say fuck anyone who argues ethics is not important when talking about efficacy. Fuck that mindset and fuck AI.

I find myself asking the same 3 questions when considering whether or not to use an AI tool.

Is it a straightforward task where I can save significant time with a tool?

Can I verify the results?

If the tool AI makes a mistake, or if I am not thorough enough in verification, is the risk of failure low?

I'm a firm believer is using the right tool for the right job. Most of the AI's people are using are not designed for the things they are using them for, and I don't see many practical uses for generative LLMs. I will acknowledge that there are purpose built AIs to be used as a tool, but even then most of them tend to do a below average job. We're also seeing a lot of algorithmic driven models being labeled as AI, similar to calling something a "smart" device.

Pretty good analogy because health damage from cars don't just come from accidents, which are terrible and tragically visible, but from pollution which is a lot less visible yet kills even more.

The problem with "LLM requires skill to use correctly." is that LLM's are not being sold as a skill but rather the end of skill.

But your analogy also breaks down because once an AI has written the code it’s very much concrete, right? The AI may choose to modify that concreteness into something differently concrete, but it’s not like you are in any doubt about what the AI comes out with (if you bother to check it, which you should).

I use to write asset systems during the AAA gaming part of my career and the elusive “[my] perfect file format” always eluded me.

Over the years I spent many a showers and long drives thinking about what the “perfect file format” needs and what kind of architecture would support this. I started formulating these into long text files and over time things crystallised into “yes this would work”.

I never had the time to write the code fully. A couple of false starts, coupled with “I should do this in rust even though I’ve been in C++ my whole life” meant it lingered and lingered.

The architecture felt right though.

Enter AI; feeding the spec text files to an AI made me resolve a couple of “ah, yes, but in practice this bit of information isn’t available yet” gotchas but the code worked. I worked in architectural layers, carefully checking input, internals and output of each before proceeding to higher layers. And tests and tests, that’s you would never have bothered to write yourself, kept each layer verifiable, even out in the far corners of reasonable.

It worked and it was real affirmation to see 20 years of pontification made real.

But then I started profiling it. And hot damn, an LLM knows nothing about cache, file systems nor does it think about ways memory allocators and virtual memory systems work.

There was 200-300% speed ups to gain from hand tuning it and iterating with a proper profiler.

And all of this happened at code level, with real concrete code.

So yeah, an AI can be a wonderful tool. And the code it produces and iterates on is concrete. It’s not like the AI starts from scratch every time you have a thought. So the very, very first pass is not deterministic. But the thing you deliver forwards very much is.

Neither does the use of an LLM mean you do not understand thoroughly what you submit. He doesn't give a crap how the code was created, as long as you, the human (I hope), understand what you are submitting and can take accountability for it. You are still responsible for whatever crap comes out of an LLM, the same way that you are responsible for what comes out of the compiler. Doesn't mean you shouldn't use compilers or LLMs.