{"slug": "people-vs-the-ai-overlords", "title": "People vs. the AI Overlords", "summary": "Debian developer and LWN editor Jonathan Corbet warns that large language models are deepening a motivation crisis in open source software, as developers increasingly rely on costly, externalized AI tools like Anthropic's Claude for code review, which he argues marks a regression to renting software and the 'enshittification' of programming. Corbet cites fellow Debian developer Russ Allbery's observation that OSS faces a motivation crisis worsened by LLMs, and highlights a typical workflow where a maintainer repeatedly runs Claude's '/code-review' command, spawning sub-agents that consume significant GPU compute.", "body_md": "# The people vs the AI overlords\n\nPreviously in this series:\n\n[The Four Horsemen of the LLM Apocalypse].\n\nIn a [post to oss-security](https://lwn.net/ml/all/87o6fmdgs6.fsf@hope.eyrie.org/), my (Debian) co-developer Russ Allbery\nstated that \"open source software [OSS] is coming face to face with a\nmotivation crisis that has been building for a long time\". His point\nis essentially that large language models (LLMs 1) are making the\nexisting OSS community crisis worse. For him, it's the flood of code\nreviews, but he argues that varies according to people's desires, for\nothers it's security issues and so on.\n\nI think Russ is right, but I would argue there's something much bigger\nthan our open *communities* going on here, and it's about the entire\n*field* of computing. This pressure is on *all* of us, regardless of\nwhether we work on open source software or not.\n\n# How people use models\n\nPeople using LLMs in their workflow have *radically* changed how\nprogramming works, even for [people who claim to avoid\nvibe-coding](https://www.spurint.org/journal/2026/07/llms-and-xfwl4). And I'm sorry to single out one poor maintainer here:\nit's not you, Brian, you're just one example among many. But this is\ntypical use of those models nowadays:\n\nOnce it’s done, I’ll use\n\n`/code-review`\n\nand let Claude spawn sub-agents to do a full review of the new code. This usually finds some problems, even problems that the “main” Claude instance didn’t find during its validation. I usually keep running`/code-review`\n\nagain and again after finding and fixing issues, until there aren’t any left.\n\nThink about what that means for a minute. This is automation built to\nfire up dozens of agents crunching at a problem for minutes if not\nhours of GPU compute time, in parallel. This is essentially a couple\nof shelves in a datacenter rack, totally maxed out on power and\ncooling, abstracted behind a cute little `/code-review`\n\ncommand.\n\nThe author, here, is rightly concerned that \"Anthropic could pull the\nrug out and require API pricing\", which is perhaps a code word for\n\"charging something closer to actual costs\". Brian also pays lip\nservice to environmental and societal costs but those are largely\nabstracted away, so let's keep that conversation aside here as well,\nas we have [discussed it before anyways](../2026-05-16-four-horsemen/).\n\nBut clearly, this way of working has an ([externalized](https://drewdevault.com/blog/Stop-externalizing-your-costs-on-me/)) cost, to\nsay the least.\n\n# Paying for non-free tools\n\nFor decades my work has been focused on free and open source software. I've long stopped using proprietary operating systems like Windows or Mac, and even before that switch, I was mostly using free software on those platforms, partly out of principle, but also because I was too poor. So the tools of my trade are free, and I build free tools with them.\n\nIt feels like we're going backwards: when I was in school, a millennia\nago, my classmates didn't have access to a compiler and were wondering\nhow they would scrape the money to buy a compiler like [Borland's](https://en.wikipedia.org/wiki/Borland_C%2B%2B)\nor [Microsoft's](https://en.wikipedia.org/wiki/Microsoft_Visual_C%2B%2B). I had a compiler built into my operating system\n([FreeBSD](https://www.freebsd.org/) at the time), so that wasn't a problem for me. For them,\nit was a significant expense, but at least those expenses (or more\n[shady sourcing of programs](https://en.wikipedia.org/wiki/Warez)) were a one-shot deal.\n\nFast forward 30 years, and software is rented: you pay monthly for\nAdobe's Photoshop and Microsoft's office suite just like you pay for\nNetflix, Disney+ or Spotify 2. And now you need to add dozens (if not\nhundreds of dollars) of monthly credits to access LLMs on top of that.\n\nSo, now we have to *pay* to get anything done? This is peak\n[enshitification](https://craphound.com/category/enshittification/) of our job: first they steal our work to train their\nmodels, and then they sell it back to us at a profit.\n\n# Attacking the engineers\n\nAI is coming for our jobs, as engineers, if not *everyone*, according\nto the narrative. For a while now, our job market has deteriorated:\nless jobs, for less pay. Lots of skilled engineers looking for work\nand finding crap jobs then still looking while working.\n\nThis is not by accident. 3 We engineers have a\n\n*lot*of power, it is not organized, but that's just a couple of unions away (\n\n[easy](https://www.ibtimes.sg/wikimedia-foundation-declines-voluntary-union-recognition-labor-dispute-grows-91377)!). Tech overlords know this, so they are attacking our profession, directly, by forcing us to train and use models that\n\n*they*can control.\n\nEven in environments where programmers are not *forced* to use LLMs,\nthe mere pressure of other people's LLM-generated work is huge. One can\nbe forced to review LLM outputs, or just peer pressured you into\nproducing more.\n\nWe're now supposed to accelerate delivery, because models can\n*presumably* do things so much better and faster. With supply chain\nsecurity becoming such a large vector that we now have [worms crawling\naround developers accounts on NPM](https://unit42.paloaltonetworks.com/chaindrop-npm-worm-analysis/), increasing the delivery cadence\nseems like a really bad idea.[4](#fn:4)\n\nThe LLM hype is part of the larger wave of cyberwar against workers, against water, against the Earth, against all the people. This is not a matter of individually \"adapting to the reality\" or personal choice, but a political, social, hard problem we need to address collectively.\n\nPreviously in this series:\n\n[The Four Horsemen of the LLM Apocalypse].\n\n-\nI again prefer the term LLM to \"AI\" because models do\n*not*possess[intelligence](https://en.wikipedia.org/wiki/Intelligence). I did use it in the title because click baiting is apparently important, but I stopped short of calling this one \"Rage Against the Machines\" because that would be the title of every blog post I have ever made.[↩](#fnref:1) -\nYes, I know that\n[Visual Studio](https://en.wikipedia.org/wiki/Visual_Studio)is kind of free now, but I wouldn't be surprised if they turn that into a rental as well, because why not.[↩](#fnref:2) -\nBeyond sabotaging the job market, Sam Altman event wants to\nsell \"\n[intelligence as a utility](https://gizmodo.com/sam-altman-says-intelligence-will-be-a-utility-and-hes-just-the-man-to-collect-the-bills-2000732953)\" something that is just a[really bad idea](https://www.psychologytoday.com/us/blog/the-digital-self/202603/intelligence-as-a-commodity)but especially shows how megalomaniac those people are.[↩](#fnref:3) -\nThis brings back memories of\n[another era](https://en.wikipedia.org/wiki/Morris_worm), walking us back decades in terms of computer security.[↩](#fnref:4)\n\nYou can use your Mastodon account to reply to this [post](https://kolektiva.social/@Anarcat/117122589723431187).\n\n[Edited .](https://gitlab.com/anarcat/anarc.at/-/commits/main/blog/2026-08-18-people-vs-ai-overlords.md)", "url": "https://wpnews.pro/news/people-vs-the-ai-overlords", "canonical_source": "https://anarc.at/blog/2026-08-18-people-vs-ai-overlords/", "published_at": "2026-08-22 22:32:42+00:00", "updated_at": "2026-08-22 22:43:23.756906+00:00", "lang": "en", "topics": ["artificial-intelligence", "large-language-models", "ai-ethics", "ai-policy"], "entities": ["Jonathan Corbet", "Russ Allbery", "Anthropic", "Claude", "Debian", "LWN", "FreeBSD", "Borland"], "alternates": {"html": "https://wpnews.pro/news/people-vs-the-ai-overlords", "markdown": "https://wpnews.pro/news/people-vs-the-ai-overlords.md", "text": "https://wpnews.pro/news/people-vs-the-ai-overlords.txt", "jsonld": "https://wpnews.pro/news/people-vs-the-ai-overlords.jsonld"}}