Are Humans Going Away in the Work We Do?

This week at Duke, we ran a hackathon with a room full of engineering students — sharp, fast, and already living inside the future. After the keynote, after we broke into circles and teams started forming, one question kept reappearing in different conversations like a recurring signal:

"Are coders going away?"

"Is programming going away?"

"Is AI about to erase software engineering?"

I've been thinking about that question for years, and I gave them the same answer I'll give you:

No. Coders aren't going away. Programming isn't going away.

What's going away — what always goes away — is the current layer of language abstraction.

That's the part most people miss, because the loudest voices in the market don't make money by being precise. They make money by being dramatic.

The Pattern We've Seen Before

If we were standing in this same room about 60 years ago, we wouldn't be talking about TypeScript versus Rust. We'd be talking about punch cards and mainframes — how to physically encode logic into a system that didn't care about your feelings or your deadlines.

Then came assembly language, and suddenly a massive shift happened: you didn't need to "program" by feeding the machine literal physical instructions. You could write symbolic commands that mapped closer to how humans think.

And yes, every punch-card "developer" who refused to evolve got cooked. Then came higher-level languages: C, C++, Java, Python, and wave after wave of abstraction. Each wave triggered the same panic:

"This makes programmers obsolete."

"This makes real engineering unnecessary."

"This makes it too easy — anyone can do it now."

But here's what never happened: We never removed the engineering layer. We never removed problem-solving, architecture, trade-offs, systems thinking, security, or reliability.

We just moved the interface. The work didn't disappear. The syntax did.

I've Used AI at the Edge — And Here's the Truth

I'm not speaking from a place of theory. Based on my OpenAI Year in Review and analytics of my use of AI — both as an individual and our mandated use of AI at BankSocial — I am, and our company is, a top 1% user of AI in the world. As someone who uses AI to the extreme, engaging with over 30 different models in some cases, I've seen the absolute peak.

And what I've seen is that we are not living in a straight-line explosion toward replacement-of-everything. We are hitting a peak that mirrors the early 2000s. Back then, Big Data and Google Search fundamentally changed how we accessed information. There was a massive leap, then a long plateau until 2023.

Welcome to the Plateau of 2026. The novelty is wearing off and the real work begins: figuring out where this actually creates durable value.

"But John… AI Companies Say They'll Replace Everyone."

Some student asked me a version of this at Duke: "What about all these companies saying they're going to replace the entirety of humanity?"

My answer is simple: Take them at their word. How else are you going to raise $20 billion without making a claim that massive? That's not a moral judgment; it's market mechanics. They are building the plumbing, and they need to convince investors that the pipes are more important than the water.

"I think the impact of large language models is a bit overhyped for the next two years, but very under-appreciated for the next ten years." — Ken Griffin

Ken Griffin has publicly criticized near-term AI hype, warning that people are overstating what these systems will do soon. I'm not quoting him to borrow authority — I'm pointing out that even serious operators in high-stakes environments see the same thing: there's real capability here, and there's a lot of narrative froth layered on top of it.

You Aren't Being Replaced — You're Being "REplaced"

When people say coding is dying, they mean typing is dying. Typing a lot of code as the primary bottleneck is dying, and it's dying fast. AI is a new interface layer. It's the end of programming as "keystrokes per outcome."

As a top 1% user, I see the shift clearly:

• The Mastered Tool: If you ignore the tech, you will be replaced out of a role.
• The High-Level Capability: If you master it, you are REplaced into a high-level capability. You aren't losing your seat; you're being upgraded to a version of yourself that can orchestrate 30 models simultaneously. The machine handles the $15-an-hour administrative "how" so you can command the $1,000-an-hour strategic "why."

So What Happens to Junior Developers?

The entry path changes. There was a time when you could learn enough syntax to be useful. You could get hired to write boilerplate. AI eats boilerplate.

The new path for junior builders becomes less about "Can you type code?" and more about:

• Can you reason about systems?
• Can you test and verify?
• Can you read code critically (including AI-generated code)?
• Can you translate ambiguity into executable clarity?

The floor rises. But the ceiling rises too.

The Great Promotion: Scaling Human Judgment

This shift isn't reserved for engineers or tech elites; it is the new baseline for nearly every professional role. Whether you are in marketing, law, finance, or operations, the "administrative tax" — the hours spent drafting emails, organizing data, or researching basic facts — is being externalized to the machine. This forces a transition from being a doer of tasks to a director of outcomes.

In this new landscape, your value is no longer measured by how much "work" you can produce in an eight-hour day, but by the quality of your judgment, your ability to spot hallucinations in the machine's output, and your skill in connecting disparate tools to solve a human problem. For most, this is the great promotion: the machine takes over the repetitive "how," finally allowing you to spend your career on the creative and strategic "why."

The Quantularity

A lot of people want this story to be about "the Singularity" — a single, centralized mind swallowing everything. I don't buy that. What we are entering is The Quantularity.

Not one mind, but many. Not one system, but interoperable systems. Not replacement, but amplification. In this world, the machine is the echo, but the human remains the source. Coders don't go away; they become conductors.

The Final Answer

The future belongs to the people who can define problems clearly, orchestrate solutions, and verify reality. AI doesn't end engineering; it ends the illusion that engineering is typing.

And honestly? That's progress.