Rewriting the role: Developers in the age of LLMs

Every few decades, a new wave of abstraction reshapes how we build software, and each time, someone declares the developer obsolete. With LLMs, we hear it again. But what if that fear is missing the point?

Last November, after my talk at BDX I/O, I wrote a six-part series exploring how the developer's role is shifting, not disappearing, in the age of AI. Here's the full journey, with links to each part.

Part I - Programmers Are Always Doomed... Until They're Not

From Fortran to Java, from punch cards to low-code, every new abstraction triggered the same panic: "This time, developers are finished." Every time, we adapted. The pattern is clear: abstraction rises, fear follows, and the craft moves up a level.

Part II - When Tools Learn, So Must We

Are we being deskilled? Not quite. Automation doesn't erase skill, it shifts it. When compilers automated machine code, we learned algorithms. When LLMs automate boilerplate, we learn critical thinking. The danger isn't forgetting how to write a for loop, it's forgetting how to think about one.

Part III - Co-Authoring with the Machine

The developer's workflow is changing. We're moving from syntax recall to intent articulation, from implementation to orchestration, from writing every line to curating systems. Coding with an LLM is like pair programming with a colleague who's brilliant, fast, and occasionally delusional. The pitfalls are real: overtrust, hallucination, and loss of mental model.

Part IV - The Developer's Journey

How do developers grow when the "easy" problems are automated? Entry-level work is vanishing, and we haven't figured out what replaces it as a rite of passage. The answer lies in new apprenticeship models: tri-programming (junior + AI + senior), guided co-creation, and mentorship that teaches why, not just how.

Part V - Teaching the Next Generation

If machines can execute, humans must learn to understand. Education needs to shift from grading output to grading process. The old assignment was "build a todo app." The new one: "use AI to build it, then find three security vulnerabilities, diagram its architecture, and redesign it for scale."

Part VI - Differently Human

LLMs can generate code. They can't generate understanding. They can't be held accountable. They can't care. Curiosity, judgment, and responsibility: these aren't things we do better than machines. They're things machines fundamentally can't experience. The future of software development isn't less human. It's differently human.

The craft continues. The tools keep evolving, but the work, turning human intent into systems that matter, remains ours.

Keep asking better questions. Keep teaching the next generation. Keep caring about what we build and why.