Am I Now an Ex-Coder? The Post-Programmer World Has Already Started
There was a time when being a programmer meant something very specific.
You understood memory.
You understood how compilers worked.
You understood what your framework was doing under the hood.
You earned your stripes.
Now, with the rise of tools like Claude Code, emerging from the wider ecosystem of frontier AI labs such as Anthropic - and the cultural shift toward what many call “vibe-coding”, we’ve crossed a structural line.
The ability to produce working software is no longer limited to formally trained engineers. It’s increasingly available to anyone who can describe what they want clearly enough.
This isn’t just a productivity story.
It’s an infrastructure story.
And most people, especially developers, are focusing on the wrong part of it.
Claude Code and the Optimism Problem
When influential engineers like Boris Cherny talk about AI-assisted development, the framing is familiar:
This is acceleration.
This is empowerment.
This is productivity.
And technically, that’s true.
Claude Code and similar systems can dramatically reduce implementation time. They compress the gap between idea and execution. They make experimentation cheaper.
But there’s a typical developer bias embedded in that optimism.
Developers tend to focus on what can be built.
They focus less on what should be built.
Less on who governs it.
Less on what happens when scale amplifies flaws.
The assumption underlying much of the optimism is that better tools simply make better engineers.
But tools that remove implementation friction don’t just accelerate engineers.
They reduce the requirement for engineers in implementation-heavy roles.
That is a structural shift, not a tooling upgrade.
When “Engineer” Stops Meaning What It Used To
For years, developer tooling improved output. Higher-level languages, open-source ecosystems and cloud platforms reduced friction.
But AI-assisted code generation is different.
This isn’t faster typing.
It’s removal of the typing.
When working APIs, UI layers and data models can be generated in minutes, implementation is no longer the scarce skill. Scarcity shifts toward:
- system design
- architectural judgement
- trade-off analysis
- governance
- commercial alignment
Writing code is becoming cheaper.
Understanding consequences is becoming more valuable.
That shift changes hiring models.
It changes team structures.
It changes what “junior” means.
What Happens to Hundreds of Thousands of Jobs?
Let’s be direct.
If one senior engineer equipped with Claude Code can generate the output that previously required multiple implementation-focused developers, organisations will rebalance their teams.
Businesses optimise for efficiency. They are not structured to preserve roles out of nostalgia.
This is economic gravity.
We’ve seen similar structural shifts before. When automobiles scaled, horses stopped being transport infrastructure and became leisure animals. Entire industries reorganised within a generation.
AI-assisted development is moving faster.
Entry-level coding roles, repetitive implementation work and execution-heavy digital roles are exposed first.
The question is not whether displacement happens.
The question is how quickly organisations act on it.
Computer Science Degrees: What Are We Teaching Now?
If AI systems can generate production-ready code from well-structured prompts, what exactly are we training students to do?
If the answer is syntax memorisation or framework familiarity, that model is ageing.
The enduring value of technical education has always been:
- computational thinking
- systems understanding
- performance trade-offs
- security implications
- failure modes
- architectural reasoning
The market rewarded velocity for the past decade.
Velocity is now commoditised.
Claude Code doesn’t eliminate computer science.
It exposes shallow computer science.
Education either pivots toward deeper systems thinking and strategic engineering, or it risks preparing graduates for a narrower slice of a shrinking implementation layer.
“It’s Just Productivity”
The common counterargument echoed across much of the developer community, is that this is simply better tooling.
Just as Stack Overflow didn’t eliminate developers, AI won’t either.
There is truth in that.
But the magnitude differs.
Previous tools accelerated engineers.
This generation reduces the number of engineers required for implementation-heavy output.
That changes economics.
And economics reshapes organisations faster than culture does.
Developer optimism tends to underestimate economic incentives.
Boards do not.
The Infrastructure Problem Nobody Is Addressing
Imagine a council saying:
“We need 500 new houses.”
They build them rapidly.
But they don’t expand roads.
They don’t upgrade utilities.
They don’t model traffic flow.
That’s where we are with AI-generated software.
Claude Code enables explosive software creation.
It does not solve governance.
We are about to see:
- a surge in internally built tools
- explosive growth in shadow IT
- increased security exposure
- technical debt created at machine speed
Frameworks still need maintainers.
Low-level languages still evolve deliberately.
Security remains a human discipline.
Scaling still requires architectural rigour.
If everyone can generate code, who governs the system?
That question is not being asked loudly enough.
Governance Is No Longer Theoretical: When Frontier AI Meets the State
This isn’t abstract.
When companies like Anthropic build frontier models, they aren’t just building productivity tools. They are building systems with implications for cybersecurity, intelligence analysis, military modelling and national infrastructure.
That naturally creates tension with institutions like the United States Department of Defense.
Not because one side is “good” and the other is “bad”.
But because their incentives are incompatible.
AI labs optimise for capability acceleration and competitive edge.
Governments optimise for control, security and strategic stability.
One moves fast.
The other moves cautiously, and with reason.
When privately developed systems can rival state-level analytical capability, the question stops being “is this useful?”
It becomes:
Who governs this?
Who audits it?
Who defines deployment boundaries?
Who is accountable when misuse happens?
This is not a distant philosophical debate.
It is happening now.
If governance at the highest levels is still being negotiated, what makes us think businesses are prepared to govern AI-generated software internally?
The gap between capability and oversight is widening.
That gap is where risk lives.
It’s Not Just Developers
If AI can draft long-form content, generate SEO strategies, build landing pages, create graphics, and analyse datasets, execution-heavy roles across digital marketing and operations face similar compression.
Execution becomes cheaper.
Judgement becomes scarcer.
If your value is tied purely to producing artefacts, you are exposed.
If your value is tied to defining the right problem and understanding second-order effects, you become more critical.
So… Am I an Ex-Coder?
In the traditional sense, yes.
I no longer define my value by writing code line-by-line.
Not because I can’t.
But because it is no longer the highest-leverage activity.
The leverage now sits in:
- framing the right problem
- designing robust systems
- aligning technical execution with business intent
- managing risk
- anticipating unintended consequences
Claude Code can generate an API.
It cannot decide whether building something is strategically wise in the first place.
That layer is becoming the real scarcity.
The Real Scarcity
We are not entering a world without programmers.
We are entering a world where programming is no longer the scarce capability.
Scarcity is shifting toward:
- systems thinkers
- technical strategists
- product architects
- governance specialists
- people who understand consequences, not just outputs
This is the real age of AI.
The builders are being automated.
The designers are becoming critical.
The question is no longer whether you can code.
The question is whether you understand what should be built - and what should never exist.