For all of human history, we’ve invented tools to make our lives easier. The spear, the wheel, the printing press, the internet—all of them created by these big brains of ours so we could use our big brains less.

AI is just one invention in a long line that has changed the way we create and consume things. Although where AI will take humanity as a whole is yet to be seen (personally, I rank its current transformative effect somewhere below the internet and above the Kindle), it’s certainly been a radical shift for the tech world. AI tools have shaken up nearly all facets of tech and software—which, some may say, is the bedrock of our modern society—and has changed the way tech workers around the world navigate their everyday tasks and workflows.

This is especially true of how we create code: in 2023, GitHub purported that, of its users, 46% of developers’ code across all languages was generated by GitHub Copilot. As of April 2026, 75% of all the new code at Google is AI-generated. Source? Their CEO’s recent blog post. Google has been pushing this for a while—they went from 25% to 75% in the span of a year and a half. And, you know…nice conveniently round numbers there, Google.

I think we can all agree that a lot of what’s being written is unusable, unsecure, broken bowls of code spaghetti, but still. The sheer number of lines created by AI is growing exponentially at a pace that the mere, mighty human developer cannot keep up with.

While I’ll take the tales of 37,000 lines a day with a grain of salt—although I applaud the tokenmaxxing—it’s safe to say that coding assistants have taken the software development world by storm. As of our 2025 Developer Survey, 84% of developers and technologists have adopted AI, and 51% of them use AI tools daily.

And look, there’s no shame in it—I’ll be the first to admit I used Gemini to research all those stats—but it does raise a few questions for me. First question: What exactly are we creating when we pump out 37,000 lines of code from a mysterious, magical black box called AI? I’m guessing not even the proselytizing CEO who generated that code could tell you.

Second question: What happens when we remove ourselves from the creation process of the things we are creating? Follow-up to that one: Can we even call something fully generated by AI our own creation?

The questions I’m asking are the same philosophical quandaries we’ve been debating amongst ourselves since the first letter was pressed into paper in 1440. Someone somewhere always seems to think new technology is the end of everything forever. And look, someday they will be right about that, but right now they’re only partially correct. Technology is the end of one particular thing: the status quo.

At large, we don’t build our own furniture or handwrite letters to people in our lives. Yet, once upon a time, if you wanted a new chair or to break up with your long distance beau, you’d need to make it and write it yourself. That was the status quo. Nowadays, we have IKEA and WhatsApp, so that status quo has become obsolete. That does not mean either of those art forms—handwriting, woodworking—has disappeared completely. It’s just that we tend to leave those low-tech, old-school crafts to the artisans of the world, who have the patience and care for it. No whittling needed here.

And that word right there, artisan, is at the very heart of our timeless philosophical issue. When technology gives us the souped-up tools to build anything we want faster than ever before, what is the point of being an artisan? Who would construct a chair slowly and by hand when you could put together hundreds of IKEA chairs in the same amount of time? Who is crafting a chair just to craft a chair? Who is writing code just to write code when an AI can build you anything you want in half the time?

The artisan vs. builder dichotomy is central to the discourse around AI tool usage happening in developer communities. Recently, Mike Swift from Major League Hacking told us that coding assistants are causing a “paradigm shift” in the world of software development, changing the way developers work, how fast they build, and how they understand their identities. “Historically, being a developer was an identity because you had to know the craft to do it,” Mike said. “That's not necessarily true anymore.” And let’s not forget, this identity-destruction was actually one of the promises of AI—now, you can be a coder, a video editor, a writer, a mathematician, or a designer, all without ever having to learn how to be one. Barring stand-up comedy, AI has lowered the barrier to entry for nearly every craft. Heck, it can even diagnose diseases, although you really should not make ChatGPT your primary care provider.

For better or for worse, we’ve created endless shortcuts for ourselves with AI. Now, for the low, low price of $20 a month, you can offload your learning, your work, your interpersonal relationships, and even your health decisions to a chatbot! This offloading is especially true for how we build software, where workers are asked to build, build, build as fast as they can. Companies don’t want artisans whose slow, meticulous work produces the finest quality products the market could ask for. No, they need people in seats now. They want builders who can put together hundreds of IKEA chairs in minutes.

And AI coding assistants are a convenient, efficient way for anyone to build at lightning speed, no craftsmanship required. But IKEA chairs aren’t built to last, and neither is AI-generated software. What’s going to happen when companies break their poorly fabricated chairs and AI-generated software?

Truthfully, I don’t think the artisan vs. builder dichotomy needs to be a dichotomy. With AI, we’ve reached yet another inflection point in our history of technical advancements, one that ends the current status quo and creates a new one—one where developers evolve into artisans who build the future of tech with speed and at scale.

Before we can talk about my proposed building-artisans and artisanal-builders, it’s important to dissect why developer identities need to evolve. But first, let me talk about myself.

I’ve never claimed to be an experienced developer, nor have I ever claimed to know much about tech at all (besides that one time during my Stack job interview where I pretended to know a lot about AI to impress my future coworkers). I’m quite loud and proud about being a bright-eyed, bushy-tailed beginner at code. Yet, with the power of AI coding tools in my hand, I’ve been able to build fully-working (okay, partially working) applications in just a few hours.

I wouldn’t call myself a builder OR an artisan in this regard (I like to identify as tech ingénue), but I know for a fact there are plenty of people who would call themselves software builders now who use vibe-coding tools. How do I know this? They put stuff like, “building the next XYZ” as their LinkedIn headers without knowing a lick of code. This is not to knock these ambitious movers and shakers—I know it takes real grit and guts to build something on your own—but their status as software builders questions how necessary it really is to know a lick of code. You, too, can be a builder of software today! But what does that mean for the builders of software from years past?

AI has made the art of coding by hand purely optional, if not entirely obsolete. But what is a developer if not someone who writes code? There’s that paradigm shift Mike Swift was talking about. Developer identities can no longer be defined in contrast to non-coder identities—it’s no longer about who can build software and who can’t. Anyone with a Claude subscription (including me) can do it. The important, identifying question then becomes, “What makes developers indispensable to the software building process?”

That indispensability comes from developers’ years of experience—their artisanry. If you can take that artisanry and build with it, you’ve got a golden ticket. I’ve experienced this myself as a writer. If AI is good at generating code, it’s even better at writing prose.

In the early days of AI, when people figured out that feeding a bot a non-specific, four-word prompt could get you back multiple paragraphs of sensible writing, suddenly everyone was a writer. People with no experience could produce content comparable to my own. So much for honing your art, huh? Being a writer was no longer about being able to write. Articulate, thoughtful paragraphs with big words and poignant metaphors had become all too commonplace. Anyone with a Claude subscription (including you) could do it. If being a good writer isn’t presupposed on writing well, what did I have to prove I was a card-carrying writer?

Well, once all the hoopla around AI-generated writing faded and people got sick and tired of the rules of threes and em dashes and contrasting statements, writing became a craft again. If you wanted to be a writer in an AI world, you couldn’t just present some pieces of writing as proof—you needed to tell a good story, one that affects and engages people. That, it turned out, was the real value of my years of artisanry. And hey, those specific qualities of my artisanry are easy to build with because, AI-assisted or not, my skills as an artisan shape what I build. Because I actually know how to craft a high-quality story with my own two hands, AI really does help me write with speed and at scale, allowing me to keep up with builders who want to move faster than my fingers can type. Because I’m willing to wield my craft in this way, I get to use my artisanry to help build things I really care about, working in tandem with fast-paced builders to write things that are real and meaningful. In other words, I’m the woodworker who helps IKEA design better chairs.

Developers have the opportunity to become that artisan asset in a builder world. As Mike Swift put it, “We need to recognize that where the value is created is in the ideas, the communication, the taste. It’s knowing what to build, why to build it, and how to get the LLM or your team to build it.” The “taste” Swift is referring to comes only from expertise and experience. It’s craftsmanship. The only people who actually know what makes a chair sturdy are the people who’ve made one. The only people who actually know how to make a codebase secure, or an application fast, or software headless are the people who know how to code.

And artisan assets will be needed once AI reaches maturity. AI-generated code is great for early-stage building, but you still need technical expertise to build something secure and reliable. Scott Hanselman, VP of Developer Community at Microsoft, asked, “Is the goal to make a prototype? Is the goal to ship a rock-solid, secure banking platform that scales to millions of people and they entrust you with their data?” And while AI-generated code works well for the former, it’s the latter that our digital world is built on. The only way to build the rock-solid, secure platforms is with knowledge and technical expertise. “A non-technical vibe coder who doesn't have the specificity [or coding knowledge] could confuse the AI and themselves and get into weird hard-coded things,” Hanselman said. “There’s a big difference between writing acute utility versus shipping Windows, or .NET, or some fundamental base-of-the-pyramid-type stuff.”

If we want to keep shipping the kind of solid, safe software Hanselman is talking about—and we do—AI-generated code just won’t cut it. But our hunger for software gnaws evermore, and it can’t be satiated at the speed code can be written by hand. But maybe we don’t have to choose between moving fast and breaking things or building slow and building well. Developers now have the tools to move fast and build well—to be an artisan who builds.

We live in a world of builders. Enterprises operate like this, on a builder’s approach, where the priorities are speed and efficiency. But to create the rock-solid software that Scott Hanselman talks about, the expertise of craftsmen—the people who actually know what makes up working and secure codebases—is only going to become more and more valuable. We saw this happen with writing and storytelling, where curation and taste became the defining factors of what is and isn’t good writing.

I imagine development won’t be far behind, especially as AI’s offer of infinite possibility for code will lead to an infinite need for code. Because builders now have infinite possibilities for what they can create for code, they’re going to continue to build things that don’t work at scale—the things that break or expose sensitive data to malicious actors. To fix those infinite number of possible software programs, they’ll need the knowledge of experts who truly understand code to come in and do the real work of making it all work.

But in a builder world, it’s important for artisans to adjust and evolve their craft to meet the times. The same way we could not ignore the printing press or wish away the wheel or turn off the internet, AI-assisted work is becoming an inevitability. But these are all just tools, ones that can make the work of artisans and experts that much easier. “Where we are in the world right now is effectively the birth of power tools,” Mike Swift said. “[Like power tools,] we have optimized the tool flow to solve the problem, to build the product, to solve whatever the end user needs.” For developers, taking these power tools in stride is essential for thriving in our new AI world.

And not all things slow, meticulous, and artisanal are lost. Perhaps the status quo has changed, but just because we invented the wheel doesn’t mean people don’t walk anymore. It’s not about being an artisan or a builder. It’s about being a little bit of both. Most likely, that artisan-builder hybrid will be the identifying differentiator of this new age of software developers. To prove my point, I wrote the first draft of this piece in a notebook with a fountain pen because it helps me connect to my writing—even as I had Gemini open on my laptop next to me.

The artisanal craft is not dead—not for writing, and certainly not for software. The only reason we can build an IKEA chair without any skills is because someone with skill designed, tested, and produced the building blocks of the chair for us. Our chairs and our software still need experienced artisans. Our code still needs developers. And the artisan developer who knows how to build will be more valuable than any AI.