So, some professor type, a Dirk Riehle, is asking if AI is killing open source. Woke up to this headline staring at me from the glowing screen, and let me tell you, it’s a hell of a question to ponder before the third cup of coffee has even started its lukewarm journey south. Killing open source? Jesus. It’s always something, isn’t it? If it’s not the cloud, it’s the latest shiny gizmo, or some venture capitalist’s fever dream. The world’s always ending, especially in this racket. Always some new god or new demon poised to send us all to hell or Valhalla, depending on which Kool-Aid you’re sipping.
This Riehle fellow, he’s got a “tl;dr Nah” on his own question, which, I gotta admit, shows a certain economy of bullshit I can appreciate. He reckons AI won’t put a bullet in open source’s already pockmarked head. Remembers twenty years ago, he says, when the suits were all flapping their gums about the cloud doing the deed. And how’d that turn out? The cloud just became another layer of complexity running on, you guessed it, a mountain of open source code. Like finding out the termites are holding up the floorboards. They ain’t friends, but they’re stuck together. Peaceful coexistence, he calls it. Right. Like me and my liver.
The new boogeyman is AI-generated code. Some magic box spits out lines of Python or whatever the hell kids are coding in these days, and suddenly nobody needs to actually think anymore. Everyone can “vibe-code their way to success.” Vibe-code. Sounds like something you’d overhear at a particularly sad Hollywood party. I vibe-code my way to the liquor store most afternoons, and the only success I find is another bottle of something to dull the edges. The idea that this vibe-coding is going to replace the tangled, sprawling, often ugly but sometimes beautiful mess of human-written open source? Color me skeptical. And hand me that ashtray.
Riehle’s point, if you squint through the academic fog, is that this AI-generated stuff will probably just lean on existing open source libraries. Like a drunk leaning on a lamppost. Doesn’t mean the lamppost is obsolete; just means it’s got another burden to bear. More use of open source, he figures. And maybe he’s right. The machines will churn out their code, and that code will call functions from libraries written by some poor bastard in Nebraska who hasn’t seen sunlight in six years. The digital equivalent of a chain gang, just with less singing.
Will developers stop contributing? He says “hardly.” Says skilled devs will still factor out reusable code, contribute it back. For fun, he says. Or, more rationally, to avoid a “maintenance nightmare.” Fun. Yeah, that’s what I call spending a weekend wrestling with someone else’s half-baked library. Fun like a root canal. But the maintenance nightmare part, that rings true. You let enough of this AI-generated, one-off crap pile up, and your codebase starts to look like one of those hoarder houses. Sooner or later, you gotta call in a priest or a bulldozer. Contributing to open source, in that light, is less about fun and more about not wanting to live in your own digital filth. A kind of desperate hygiene.
And will open source developers be discouraged? “Hardly,” Riehle says again, bless his optimistic heart. AI can aid any developer, speed things up. Sure, it can. Like giving a toddler a power drill. It’ll speed things up, alright. The question is, speed up what? The creation of elegant solutions, or the proliferation of new and exciting ways to screw things up? My money’s usually on the latter. Humans have a genius for that, and I see no reason AI won’t just amplify the talent. “Quality tools in the hands of skilled developers,” he says. Yeah, and a bottle of bourbon is a quality tool in my hands, but it doesn’t always lead to high-quality prose. Sometimes it just leads to a nap on the floor.
The idea that users will suddenly start cooking up their own applications from scratch because some AI can burp out a few functions? That seems about as likely as me swearing off cheap whiskey. Most people can barely work their goddamn toasters. They’ll build on open source, or pay some other poor schmuck to build on it for them. The food chain remains intact. The big fish eat the little fish, and the AI just makes the water murkier.
Now, Riehle does see a couple of snakes in this garden of digital Eden. One is the “down (dumb?) skilling of developers.” The other is the “unresolved copyright issues.” Let’s take the copyright mess first. That’s always a good time. AI trained on mountains of open source code – who owns the output? It’s like trying to figure out who owns the vomit after a frat party. Good luck with that. Riehle says, “For now you can safely assume that everything is AGPL-3.0 licensed, you just can’t prove it.” I love that. It’s got the authentic stink of the streets. It’s yours, but don’t get caught holding it. The lawyers, of course, are probably rubbing their hands together so hard they’re starting fires. More billable hours, more confusion, more ways to stick it to the little guy. Business as usual. Another cigarette tastes like ash and regret, but what else is new.
Then there’s the down-skilling. Or as I like to call it, the acceleration of incompetence. “Some people will respond to the challenge and apply themselves and learn and get better. And some people are just lazy.” No shit, Sherlock. That’s been the story of humanity since the first ape decided it was easier to club his buddy and steal his banana than pick his own. AI code generators, Riehle figures, might get more people “into the game.” Swell. More cooks in a kitchen that’s already on fire. He admits this “creates a broader distribution and a longer tail of poorly written code, now not just available in Excel but also in Python on GitHub.” That, my friends, is poetry. We’re democratizing shitty code. Making it accessible to everyone. What a triumph for progress. Users will have to “adapt to this changed reality.” Adapt by developing a higher tolerance for software that crashes, burns, and generally behaves like a meth-addled honey badger. The future’s bright, ain’t it?
The core of it, Riehle says, is that these AI code-farting machines will assist developers. And whether that assistance is used for “one-off (throwaway) code or turned into high-quality open source code is up to the developer.” There’s that optimism again. Like believing the guy at the end of the bar who says he’s only having one more. The developer. The poor, beleaguered, often underpaid and over-caffeinated developer. They’re the bottleneck, he says. Past, today, and future. The skilled ones, anyway. They’re the ones who have to sift through the digital detritus, the AI droppings, the legacy spaghetti, and try to make something that doesn’t immediately self-destruct. They’re not just developers; they’re sanitation engineers for the information age. And their job just got a whole lot smellier.
I light another cigarette. The smoke curls up, dances with the dust motes in the early morning gloom that passes for light in my apartment. This Riehle, he’s not wrong, not entirely. Open source isn’t going to just keel over and die because some glorified autocomplete can write a sorting algorithm. It’s too stubborn, too entrenched. It’s like a persistent cough, or a bad habit you can’t quite kick. It’s useful, in its chaotic, sprawling way. It’s got that human touch, which means it’s flawed, messy, and occasionally brilliant.
AI is just another tool. A powerful one, maybe. But so is a hammer, and you can use that to build a house or bash someone’s brains in. The potential for fuck-ups is enormous. Imagine, an AI diligently trained on all the open source code out there – all the security holes, all the outdated practices, all the weird compromises made at 3 AM by some developer fueled by stale pizza and desperation. And then this AI starts replicating those patterns at scale, at lightning speed. It’s like arming an army of well-meaning but catastrophically stupid robots. What could go wrong?
The “collaboratively developed high quality broadly usable software” Riehle talks about – that’s the ideal. The reality is often a bunch of well-intentioned people pulling in slightly different directions, held together by duct tape, caffeine, and a shared sense of mild panic. And you know what? Maybe that’s its strength. It’s adaptable because it’s always a bit broken. It’s resilient because it expects to get punched in the mouth.
AI might help sharpen its profile, Riehle suggests. Maybe. Or maybe it just provides a new, slicker way to generate even more code that needs fixing, more problems for the human developers to solve. The “skilled developer” remains the hero, or the victim, of this story. The one who has to understand not just the code they wrote, but the code the machine wrote, and why the machine wrote it that way, and how to fix it when the machine, inevitably, screws the pooch. It’s a hell of a job description. Needs a good dental plan and a lifetime supply of antacids. And whiskey. Lots of whiskey.
The cloud didn’t kill open source, no. It just put it to work in bigger, hotter server farms. AI won’t kill it either. It’ll just… change the flavor of the chaos. We’ll have AI-assisted bugs, AI-hallucinated features, AI-plagiarized vulnerabilities. It’ll be a brave new mess. And the open source community, bless its ragged, anarchic heart, will probably just sigh, roll up its sleeves, and start patching. Because that’s what humans do. We make messes, and then we try to clean them up, or at least make them slightly less catastrophic.
So, yeah, Riehle’s probably right. Open source will lumber on, like a cockroach after a nuclear war, too stubborn and too widely dispersed to kill. AI will be another tool in the shed, sometimes useful, sometimes just another way to hit your thumb. The real question isn’t whether AI will kill open source, but whether the combination will finally drive the remaining sane developers into hermitage or heavy drinking. My bet’s on the latter. At least there’s some solace there.
The sun’s probably thinking about coming up. Or maybe that’s just the streetlight. Time for another cup of something that tastes vaguely of regret and ambition. The world keeps spinning, the code keeps piling up, and the hangover’s always waiting in the wings.
Keep your heads down, and your glasses full.
Chinaski.
Source: Is AI killing open source?