In The New York Times Magazine, writer Clive Thompson‘s sweeping piece “Coding After Coders: The End of Computer Programming as We Know It” asks a question that is keeping a lot of software engineers up at night: what actually happens to programmers when AI can do most of the programming? Thompson reportedly spent months speaking to more than 70 developers at Google, Amazon, Microsoft, Apple, and beyond to find out. The picture that emerges is complicated, surprisingly optimistic in places, and quietly unsettling in others.

Below are the key threads running through the piece:

• The article opens against the backdrop of AI coding tools like Claude and ChatGPT becoming genuinely capable of writing, debugging, and deploying code autonomously. Engineers are watching these agents handle tasks that used to take weeks, and many are scrambling to figure out where that leaves them.
• Most developers interviewed landed on the optimistic side of the debate. The historical comparison that kept coming up was the compiler, a tool from the 1950s that many feared would end programming as a craft. Instead, it lowered the barriers to entry and the industry exploded. Many believe the same dynamic, known as the Jevons paradox, could play out again here.
• One reason programmers may have an edge over other professions facing AI disruption is testability. As Simon Willison, a tech entrepreneur and influential AI blogger, put it: “I feel like programmers have it easy. If you’re a lawyer, you’re screwed, right?” Code can be run to see if it works. A legal brief written by an AI has no equivalent safety net before it reaches a judge.
• Not everyone is comfortable with the shift. One Apple engineer, who asked to remain anonymous to avoid professional blowback, pushed back on the enthusiasm: “I believe that it can be fun and fulfilling and engaging, and having the computer do it for you strips you of that.” The fact that he felt he needed to stay anonymous says something about how corporate culture is shaping this conversation.
• Beneath the optimism, a real anxiety runs through the piece. Some engineers worry about what has been called the “permanent underclass” problem, the fear that a generation of workers once promised stable, well-paying tech careers could find themselves structurally left behind as the skills that got them hired become commoditized overnight.
• The article also touches on autonomy and risk. AI coding agents can now push code directly into production environments. Developers argue this is manageable precisely because software is verifiable in ways that other AI-generated work is not, but it still represents a significant shift in how much humans remain in the loop.

Thompson’s piece resists easy conclusions. The developers who come across as most confident are those who see AI as expanding what they can build rather than replacing what they do. But the piece is honest that this framing may not hold for everyone, and that the transition will be uneven. The compiler analogy is comforting. Whether it actually applies to a tool this capable remains, for now, an open question.

Note: This post was drafted with the assistance of Claude AI, drawing on reporting from the original New York Times Magazine article and related commentary. Refer to the original sources below for full context.

Related Resource links

• Coding After Coders | The New York Times Magazine
• Simon Willison’s commentary on the piece