Quoting Charity Majors: AI demands more engineering discipline, not less

Origin: a tweet that struck a nerve Sometimes a sharp one‑liner does more to spark real debate than a long technical blog post. Charity Majors, co‑founder of Honeycomb, wrote a few sentences that were widely shared by Simon Willison and others, shaking up the developer community: “What happened in 2025 was this: the economics of code production were turned upside down. Instead of being very hard, time‑consuming, and expensive to generate code, it became effectively free and instant. Lines of code went from being treasured, reused, cared for and carefully curated, to being disposable and regenerable, practically overnight.”

This resonates because it captures what every developer using AI coding tools feels: code comes too easily now. But what comes next?

The hidden cost of free code At first glance, zero‑cost code seems like a utopia. However, Majors’ real point is that it demands “more engineering discipline, not less.” When code can be generated at will, it loses its old filtering function. Previously, writing a feature required careful thought because high cost naturally weeded out sloppy approaches; now you can generate ten versions in an instant, but judging which one is correct, maintainable, and secure becomes a purely human burden.

Consider an analogy: a 3D printer makes manufacturing easy, but designing a tool that is both useful and safe still requires a professional engineer. AI lowers the “fabrication” barrier but raises the bar for “design” and “verification.” Code becomes throwaway paper, yet system architecture, data flows, and security boundaries—things you can’t casually throw away—become even more crucial.

The shifting value of a developer This isn’t new. Every wave of automation, from assembly languages to cloud, devalues certain skills while elevating higher‑order abilities. AI‑assisted programming isn’t “replacing developers”; it’s redefining them from code‑writing craftspeople into system‑level orchestrators.

In the future, a great engineer might be measured not by lines written but by lines wisely deleted or by how deftly they compose AI‑generated modules. For teams, that means hiring criteria must change, career paths must adapt, and code review practices must evolve.

What you can think and do right now

• As a developer, don’t mistake speed for “think less, write more.” AI helps you write faster, but you should invest even more time in design, validation, and reviewing what the AI produces.
• On teams, strengthen code review culture and extend it from “finding bugs” to “does the design decision make sense?” When code may never have passed through a human mind, design flaws hide easily.
• If you manage projects, recognize that technical debt will accumulate faster and more subtly. The slow creep of debt can now be buried almost overnight by single‑use AI patches.

The surprising twist Many assume AI will make junior developers obsolete first. Majors’ insight suggests a different pattern: junior devs may rapidly harness AI and boost productivity, while senior engineers’ real moat lies in understanding complex system boundaries and taking ownership of correctness—things AI struggles with. Yet the irony is that when code becomes disposable, organizations may let architecture rot under the illusion that “we can always regenerate it,” until the system collapses like a too‑tall tower of LEGO bricks.