Most of the solarpunk crowd seems to equate anything LLM with Sam Altman and Elon Musk. They think it is a purely capitalistic endeavor that can’t run on anything else than methane-breathing datacenters. There needs to be some education about the real impact of it and the open source of it. To explain how it can fit into a post-capitalist society.

I do think that vibe-coding is one way to reappropriate tech yes, and is extremely solarpunk. It makes manipulating machines and designing system a far more inclusive capability, bringing it from the work of specialist into the political sphere.

But explaining that is an uphill battle. When I made a post about solarpunk AI a year ago, it was well received. I fear it would be downvoted into oblivion if I published the same thing today.

The concept is new to me, so I’m a bit challenged to give an opinion. I will try however.

In some systems, software can be isolated from the real world in a nice sandbox with no unexpected inputs. If a clear way of expressing what one really wants is available, and more convenient than a programming language, I believe a well-trained and self-critical AI (capable of estimating its probability of success at a task) will be highly qualified to write that kind of software, and tell when things are doubtful.

The coder may not understand the code, though, which is something I find politically unacceptable. I don’t want a society where people don’t understand how their systems work.

It could even contain a logic bomb and nobody would know. Even the AI which wrote it may tomorrow fail to understand it, after the software has become sufficiently unique through customization. So, there’s a risk that the software lacks even a single qualified maintainer.

Meanwhile some software is mission critical - if it fails, something irreversible happens in the real world. This kind of software usually must be understood by several people. New people must be capable of coming to understand it through review. They must be able to predict its limitations, give specifications for each subsystem and build testing routines to detect introduction of errors.

Mission critical software typically has a close relationship with hardware. It typically has sensors coming from the real world and effectors changing the real world. Testing it resembles doing electronical and physical experiments. The system may have undescribed properties that an AI cannot be informed about. It may be impossible to code successfully without actually doing those experiments, finding out the limitations and quirks of hardware, and thus it may be impossible for an AI to build from a prompt.

I’m currently building a drone system and I’m up to my neck in undocumented hardware interactions, but even a heating controller will encounter some. I don’t think people will experience success in the near future with letting an AI build such systems for them. In principle it can. In principle, you can let an AI teach a robot dog to walk, and it will take only a few hours. But this will likely require giving it control of said robot dog, letting it run experiments and learn from outcomes. Which may take a week, while writing the code might have also taken a week. In the end, one code base will be maintainable, the other likely not.

My knowledge is very limited in coding and since this is the first time I hear the term vibecoding, I don’t think I can answer your question just by reading the wiki you linked. Don’t get me wrong, I think it’s great you did link it!

So I thought of sharing one myself. Perhaps it could help you make up your mind on how to answer your question? I dunno, I suppose at least, it could be a good starting point, and I hope you totally enjoy reading it!

A Solarpunk Manifesto

(warning: I hate “vibe” coding for a lot of reasons, and even more what it represents)

LLMs are the opposite of anything ecological IMHO.

What if you could build just the habit-tracking app you need

We have a thousand of those already. A better example is needed.

mold an existing app

That’s not how any of this works. One more reason to shun those who do not care and take the time to understand what programming is all about.

the capital requirements of software development from millions of dollars

Linux is free FFS, install Ubuntu today and you have all the languages you’ll ever need. How is code vomit vibe coding helping? Also LLMs are very expensive to run right now, it’s the worst example.

Last but not least, I hate how all the CEOs, managers, companies, and random people try to: pretend that open-source does not exist, change the meaning of the word open-source by associating it with binary blobs, and show developers as selfish people (“tech wizards”) who want to keep the technology for themselves.

You don’t want to learn how computer works and it’s fine, it’s your right, but don’t pretend it’s anyone’s fault.

using insecure code that a glorified autocorrect has spat out hopefully isn’t going to be a part of the future I’ll be living in.

I think the pretty universal answer in all these comments is “no”- I think that’s fair but I’d add sone caveats.

There’s a lot of negative sentiments here around LLMs, which I agree with, but I think it’s easy to imagine some hypothetical future where LLMs existing without the current water/energy overuse, hallucinations or big companies stealing individuals work. Whether that future is likely or not, I think it’s possible.

The main reason vibe coding isn’t solarpunk is that, taken by itself, it’s not in any way related to ecological stewardship, anti-capitalist community building, or anything else that’s core to solarpunk. Vibe coding might or might not be part of some “cool techy future” in the same way as flying cars, robots, and floating cities but that’s not a reason to consider it as solarpunk.

If you’re into LLMs and solarpunk, instead of arguing that LLMs are solarpunk, you can make efforts to push them to being more solarpunk. How can LLMs support communities instead of coorporations? How can, through weights sharing and various optimisations, we make LLMs less damaging to the environment? Etc. That’d at least be a solarpunk way to go about LLMs, even if LLMs aren’t inherently solarpunk.

If you want to advance humanity through free libre software, look at the FLOSS movement; that’s kinda their whole thing. Releasing a small piece of software on GitHub and providing some decent documentation on it is a nice thing to do.

Also, yeah, programming with an llm can speed things up, but you have to know enough to recognize when the llm is hampering you and you have to just roll up your sleeves and code the damn thing yourself. They’re improving, but they are still kinda stupid and they lie.