Technology

Still looks like they allow it as long as it gets approval by the core team. Not saying I agree with that choice, but there's still a very real chance LLM-generated code makes it into NetBSD.

Code generated by a large language model or similar technology, such as GitHub/Microsoft's Copilot, OpenAI's ChatGPT, or Facebook/Meta's Code Llama, is presumed to be tainted code, and must not be committed without prior written approval by core.

Pretty wild seeing a programmer anthropomorphize their applications and mislabel them as "AI".

The Big Lai runs deep.

Whatever's the outcome, the anthropomorphization in that message stinks 🤮

Hackers use LLMs to spot odd exploits so it makes sense developers use them too. The usage of LLMs in development likely won‘t make Linux better or safer overall, because bugs and exploits wouldn‘t have been found by either side without it in the first place. It‘s like pandora‘s box where Linus is kind of forced to use it. If that makes sense.

"People are going to murder anyway, so why even make murder illegal?"

You are correct, you should make rules even understanding that they will be broken by some.

I hate this attitude that if a rule or standard can’t be enforced perfectly, we shouldn’t strive for it.

"Murder bans may be impossible to enforce in the first place"

This isn't a case where something can't be enforced perfectly, it is a case of something that can't be enforced at all going into the future. Every tool that can reliably detect LLM code is at the same time the tool used for adversarial training, making the generated code look more and more human. We are already at the point where for plain english the false positives and false negatives go through the roof, making these tools very unreliable and when applied automatically a liability. Code is a lot more formalized, with a lot less personal variance (spelling, vocabulary and grammar are basically fixed - only the used logic and how it is implemented is variable), making detection harder by default than in natural languages.

If LLM code can't be detected anymore by automated means - and that state of things is approaching fast - then any policy about allowing or restricting LLM code is not worth the paper you would use to print it out. But that's not so much of a problem. The more important policy to set, that can also be enforced, is that everyone submitting code has to take personal responsibility regarding the quality of the submission. Delivering bad code - when not happening while training to become a better coder and looking for feedback - has to lead to consequences based on the seriousness of the case and if it's a repeat offender. Anyone using an LLM to spit out bullshit LQ code will run into that kind of rule very fast.

I remember people being angry at IDEs for making programming too easy and belittling code colourization as for children.

There will always be a crowd that wants everything to stay the way they know it.

Any engineer fighting the use of LLM’s will ultimately find that their peers are achieving more in a shorter time then they are.

How do you measure that?

I am asking because I tried all the AI coding tools (copilot, cursor, claude code, etc.), even the paid versions, to find the utopia that doubles my productivity where lots of people are talking about. But everytime I find they take more time to generate code than me typing it, they rarely get things right the first time, they produce way more code than needed (which is much more to reason about) and all in all it takes longer to achieve something using "AI" instead of doing it on my own.

The worst results are the ones that seem to be correct, but can only be found incorrect on closer inspection. For instances, a part of my job is to write data collection applications for sensors; those sensors come with a datasheet (usually 100 to 200 pages) that explain how to "talk" to the sensor. Sounds perfect for an LLM - throw the datasheet in, get a small C application out. But either I get something that does not even compile, or something that seems to collect data, but configures the sensor completely wrong. Debugging that takes more time than just working through the datasheet myself and program it myself.

ethical questions aside

And his here is the problem. Oh yeah lets just waive off the biggest critisicm on and issue of AI.

Hitler was a pretty neat guy, ethnic cleansing aside.

And before you say this, no, i am not comparing you to hitler, I'm only using an over the top example to hopefully make my point more clear.

Yeah, ethical, environmental, legal and fuck the consumer memory questions aside.