Interview: Drew DeVault on an AI-free version of Vim
Technology, Environment, and Art
SubscribeSign in
Interview: Drew DeVault on an AI-free version of Vim<br>Why one programmer is not happy with AI
Dr. Jason Polak<br>Jul 07, 2026
Share
I’m pleased to present an interview with Drew DeVault, a programmer who recently forked the Vim text editor as “Vim Classic”. He did this to make an AI-free version of Vim, the classic Unix text editor developed by the late Bram Moolenaar. Bram developed Vim as an evolution of the original Vi.
The Vim Classic website states, “Vim Classic is a fork of Vim 8.x for long-term maintenance, providing a stable, dependable editor — maintained entirely by humans.”<br>In particular, it doesn’t use LLM-assisted code. In an age where most programmers are embracing AI, this made me curious. I found some hints to Drew’s thinking on his blog, where he says:<br>GenAI is something I care about. It causes a lot of problems for a lot of people. It drives rising energy prices in poor communities, disrupts wildlife and fresh water supplies, increases pollution, and stresses global supply chains. It re-enforces the horrible, dangerous working conditions that miners in many African countries are enduring to supply rare metals like Cobalt for the billions of new chips that this boom demands. And at a moment when the climate demands immediate action to reduce our footprint on this planet, the AI boom is driving data centers to consume a full 1.5% of the world’s total energy production in order to eliminate jobs and replace them with a robot that lies.
Let’s get on with the interview to find out more!<br>The interview
Jason: So, you released a new version of Vim called “Vim Classic”, which is, as is stated on the project website, “maintained without the assistance of generative AI tools”. First of all, I think that’s awesome because Vim was one of the first text-editors I learned to use...that is, after I figured out how to exit the editor. Haha. But seriously, I really like Vim, and how mastering it leads to more efficient text editing.<br>I also noticed on your blog that you have a few gripes with generative AI. What I’d like to know is, for you, was creating an AI-free fork of Vim more of a practical matter (i.e. maybe hand-crafted software without AI leads to better software, or software that’s easier to maintain without all those AI pull requests, etc.) or is it more of a philosophical matter relating to how AI is changing the world?
Drew: First of all, I can tell you that we have no intention of making it easier to exit Vim, so you and your readership can rest assured on that point.<br>Choosing not to use generative AI tools, and organize projects that reject it, is both practical and philosophical, and also more: it’s an ethical and political imperative, and moreover it is a matter of self-preservation.<br>In practical terms, there are many reasons to reject LLMs. You noted that “sloppy” pull requests are a big problem for projects who don’t reject LLM-assisted contributions, and you’re right. A pragmatic concern which is even more important is the importance of copyright and establishing the provenance of your software — an important job for any FOSS maintainer who accepts public contributions, and one which is utterly impossible when LLMs are involved. Furthermore, it’s been established that the use of LLMs causes “deskilling”, making its users dumber and less competent over time, a fate I have no interest in myself.<br>In philosophical terms, using AI subverts the craft of software engineering. The hardest part has never been writing code, but knowing what code to write. Software engineering has never been taken seriously as a practice — people have been looking for shortcuts and quick, messy answers for as long as any of us have been doing the job, a sort of “move fast and introduce complexity” approach which leaves excellence out of the equation entirely and leads to things like a micro-dependency culture and all of its problems.<br>But I have long rejected and advocated against this approach, in favor of a discipline which more resembles the “engineering” from which it takes its name: an understanding of your problem domain, careful thought, reflection, and planning, and executing on the solution within the constraints given. LLMs are the most grotesque rejection of this discipline we’ve seen to date, and it exponentially accelerates the pace at which software can be made even shittier than it already is.<br>Then there’s the ethical arguments, which are myriad and well-known. The “well-known” fact led me recently to characterize AI users and apologists as “good Germans” (someone emailed me after I said that, actually, and told me it was what convinced them to ditch AI). I feel it’s pointless to rehash the ethical problems that everyone surely ought to understand by now, given that they seem to fall on deaf, cowardly ears, but nonetheless:<br>The environmental impact of generative AI is astronomical and only...