I’ve been trying to put my finger on what feels fundamentally different lately. Not the obvious stuff—yes, we have AI assistants now. But it’s deeper than a new monthly subscription.

We’re at an odd inflection point where each of us has access to a coding machine that, in most practical senses, is the best coder in any room we walk into. That’s not hype; it’s just the situation we’re in. So what happens when the baseline technical bar moves because of something that isn’t human?

The ladder just got shorter

For most of my career, the progression felt predictable. Junior developers could write functioning code but needed guidance. Mid-level developers could deliver features independently. Senior developers could design systems and anticipate problems before they happened.

But the real differentiators as you moved up weren’t raw coding speed or syntax knowledge. They were softer skills, earned through experience: knowing when to zoom out, communicating tradeoffs, designing processes that worked for people, making the team better instead of just the codebase. Those things took years to develop, which is why they defined seniority.

Now the early-career curve is shorter. AI handles much of what juniors used to struggle with. If a junior can generate decent code with the right prompts, what are they actually learning in those first couple of years? Prompt engineering? And if that experience gets compressed, the differentiators become the same skills that used to sit at the top of the ladder.

You can see it in the job market: fewer junior roles, higher expectations for the ones that exist, “entry-level” postings that quietly assume two or three years of experience and strong communication skills. None of this is malicious, but it makes starting a career in coding harder than it used to be.

Craftsmanship vs. production

Where do you find satisfaction in the work now? What actually gives you that sense of “this was worth my day”?

For me, coding was always a means to an end. I like building things people can use. The typing-into-the-editor part was never the point—it was just the most effective tool I had. So professionally, this shift doesn’t bother me as much. AI shortens the distance between idea and outcome.

But I do miss the puzzle-solving. The elegance of a thoughtful solution. That dopamine hit when something difficult finally works. That feeling changes when the code arrives from a prompt. The work can feel cheap and soulless when the process disappears and only the outcome remains.

The question everyone’s thinking about

Are developers being replaced? Some will be. But the more interesting questions are less obvious: How many developers do we actually need? What does the job become when writing code is ubiquitous? And how do we grow the next generation when the old path no longer exists?

We learned by doing—by breaking things, by debugging production incidents. That kind of judgment doesn’t come from prompts. Despite the ongoing efforts to replace us, someone still has to run the code machine. Someone has to read what the AI produced and decide whether it makes sense, understand why it chose that approach, determine whether it’s solving the right problem, and figure out what happens when it breaks in production. Someone has to maintain and refactor a growing pile of generated code that appeared in seconds and will live for years.

We’re expected to be project managers, requirements translators, and calm explainers of technical decisions, because those are the skills required to operate the machine effectively. If AI short-circuits the early years of a developer’s career, the industry may struggle to produce people who have those skills.

Historical comparisons

Looking for historical comparisons, I think about mathematicians when computers became popular. Before digital computers, complex calculations still needed to be done. Finance, engineering, and physics all relied on computation. We addressed this by hiring human computers—people whose full-time job was to do calculations by hand.

Then mathematicians created machines that removed the need for human calculators. At the time, many feared the mathematics industry might shrink. In reality, the opposite occurred. Once calculation was no longer a bottleneck, the field shifted toward concepts, models, and questions that had been impossible to explore when every result took weeks. Mathematics became more creative, more powerful, and more essential. The human computers disappeared because we no longer needed them. Now we all carry calculators in our pockets. While this hurt some people, the industry didn’t die.

I think about this a lot when people wonder if AI is replacing developers.

Agency over efficiency

As you think about where you fit in all this, one thing feels solid to me: don’t become the human computer. Human computers were very good at their jobs, and they were immediately replaceable once a better tool existed. The people who built the machines were fine because they understood the system, the concepts, and the reasons behind the work.

Agency matters more than efficiency now—understanding when to use tools, when not to, and how to explain decisions to other humans. Designing systems. Making judgment calls. Those skills don’t scale as well as code generation, which is exactly why they remain valuable.

Technical ability still matters, but it’s no longer sufficient on its own. We had a long stretch where raw technical skill could carry you pretty far, and that stretch feels like it’s coming to an end. Maybe that’s how it should’ve been all along.

What matters when code is cheap

Judgment, context, and taste. Knowing what not to build. Recognizing when something technically works but strategically fails. Being able to read generated code and say “this will hurt us later” and explain why.

As this evolves, the lines between developers, project managers, and requirements translators blur. Those roles don’t disappear; they become more integrated. Someone still needs to turn wants into action and resolve tradeoffs, and more often that someone is the developer. AI alters careers in challenging ways—expectations rise sooner, and fewer people get the chance to learn quietly.

Writing code remains essential, but it’s no longer the leading indicator of value. The people who endure will be those who can understand the entire problem, use the tools thoughtfully, and recognize when the technology is outpacing them.

Personal stakes

For most of my career, my ideas were bigger than my ability. I’d think of something ambitious, then get lost in the details of implementation, or fail to communicate my vision clearly enough to gain momentum, or move on to the next thing.

AI removes some of that friction for me. It helps me research, organize ideas, and quickly create rough versions so I can show people something real. The gap between concept and prototype has narrowed, and that feels both exciting and unsettling.

If nothing else, I don’t want to look back and see that I spent this transition trying to turn myself into a human computer.