Using that term through the article makes it hard to take seriously. I know nothing of this project but right off the bat it seems like a project that has little credibility just because of the tone used throughout.
There's no need to turn it into a full-on tirade against this set of technologies either. Is this an appropriate place to complain about Reddit comments?
Ironically, the author could well benefit from running this slop through an llm to make it more professional.
the impact of increased energy consumption, from non-zero-emission sources, on global warming is a highly practical matter based on established science
AI is/will cause a significant increase in energy consumption (still largely powered by fossil fuels) at a time when it's well established that we're supposed to be reducing emissions (Paris Accord etc.)
The IP point: this hasn't yet been fully resolved, and I expect companies to continue to fudge it by sticking "if disney return false" in their image generators. My own employer has a stance of "no AI generated code in production use" (but allowed for testing and infrastructure which will not be distributed).
I'm suprised they didn't make a point which is especially painful for Open Source projects, that AI might reduce coding effort but increases review effort, and maintainers are generally spending the majority of their time on review anyway. Making it easier to generate bad pull requests is seen as well-poisoning.
> "These resources are better used on quite literally anything else."
What shocks me most is that we have found something less useful than bitcoin mining. Remember all the articles about the environmental impact of bitcoin? That is peanuts compared to what the worlds largest companies are biulding to power the next LLM.
This is very hard to take seriously, it feels ideologically driven.
The introduction where they claim LLMs are useless for software engineering is just incorrect. They are useful for many software engineering tasks. I do think that vibe coding is rubbish however, and more junior SWEs very regularly misuse LLMs to produce nonsense code.
The only substantive point is that the LLM may regurgitate pieces of proprietary training data; although it seems unlikely that it would be incorporated wholesale into the codebase in such as way it matters or opens them up to liability.
I do question if LLMs would even be useful for such a niche project -- but I think this should be left up to developers to figure out how it complements their workflows rather then ruling out all uses of LLMs.
EDIT: I want to point out that I think the Asahi Linux project is a jewel of engineering and is extremely impressive.
In the future there will be safe havens where LLM generated code has not been merged. It will marketed as “hand-crafted” by Romanian programmers or something like that, akin to Swiss watches. It will be extremely high quality, but too expensive to mass produce.
Agreed on the IP point. Strongly agreed on not wanting to go to court in the current US political climate.
On the rest, especially the confidently incorrect argument.. Not. so. much.
Firstly, models are stochastic parrots, but that truth is irrelevant because they're useful stochastic parrots.
Second, hallucinations and confidently incorrect outputs may yet be a thing of the past, so we should keep an open mind. It's possible that mechanistic interpretability research (a fancy term for "understanding what the model is thinking as it produces your response") will allow the UI to warn the user that uncertainty was detected in the model's response.
Unfortunately none of that matters because the IP point is a blocker. Bummer.
It seems apparent in this thread that any remote criticism of AI results in downvotes.
Yes, the article might have some wording issues, but for an operating system project to choose to not allow AI written code for a product that is inherently in need of good security, and rather opt for “think before you write and fully understand what you are doing”, I don’t think that their choice is invalid.
I wouldn’t wanna get into a plane where have the core systems are written by hallucinating AI.
Have whatever opinion you want , we're going to rapidly run into issues where certain organizations are outran by others that have more liberal policies about this kind of work.
Call it slop all you want, doesn't change the unreasonable effectiveness that some individuals seem to have with such systems.
19 comments
[ 5.4 ms ] story [ 31.5 ms ] threadUsing that term through the article makes it hard to take seriously. I know nothing of this project but right off the bat it seems like a project that has little credibility just because of the tone used throughout.
There's no need to turn it into a full-on tirade against this set of technologies either. Is this an appropriate place to complain about Reddit comments?
Ironically, the author could well benefit from running this slop through an llm to make it more professional.
how so?
the impact of increased energy consumption, from non-zero-emission sources, on global warming is a highly practical matter based on established science
AI is/will cause a significant increase in energy consumption (still largely powered by fossil fuels) at a time when it's well established that we're supposed to be reducing emissions (Paris Accord etc.)
There’s no proof nor counter-proof that human brain doesn’t work like that.
I'm suprised they didn't make a point which is especially painful for Open Source projects, that AI might reduce coding effort but increases review effort, and maintainers are generally spending the majority of their time on review anyway. Making it easier to generate bad pull requests is seen as well-poisoning.
What shocks me most is that we have found something less useful than bitcoin mining. Remember all the articles about the environmental impact of bitcoin? That is peanuts compared to what the worlds largest companies are biulding to power the next LLM.
>FOSS projects like Asahi Linux cannot afford costly intellectual property lawsuits in US courts
seems quite practical and non Ludditeish.
The introduction where they claim LLMs are useless for software engineering is just incorrect. They are useful for many software engineering tasks. I do think that vibe coding is rubbish however, and more junior SWEs very regularly misuse LLMs to produce nonsense code.
The only substantive point is that the LLM may regurgitate pieces of proprietary training data; although it seems unlikely that it would be incorporated wholesale into the codebase in such as way it matters or opens them up to liability.
I do question if LLMs would even be useful for such a niche project -- but I think this should be left up to developers to figure out how it complements their workflows rather then ruling out all uses of LLMs.
EDIT: I want to point out that I think the Asahi Linux project is a jewel of engineering and is extremely impressive.
And yet, weather prediction works. Therefore, LLMs work?
On the rest, especially the confidently incorrect argument.. Not. so. much.
Firstly, models are stochastic parrots, but that truth is irrelevant because they're useful stochastic parrots.
Second, hallucinations and confidently incorrect outputs may yet be a thing of the past, so we should keep an open mind. It's possible that mechanistic interpretability research (a fancy term for "understanding what the model is thinking as it produces your response") will allow the UI to warn the user that uncertainty was detected in the model's response.
Unfortunately none of that matters because the IP point is a blocker. Bummer.
Yes, the article might have some wording issues, but for an operating system project to choose to not allow AI written code for a product that is inherently in need of good security, and rather opt for “think before you write and fully understand what you are doing”, I don’t think that their choice is invalid.
I wouldn’t wanna get into a plane where have the core systems are written by hallucinating AI.
Call it slop all you want, doesn't change the unreasonable effectiveness that some individuals seem to have with such systems.