A Dissent: English Is Not the New Programming Language

In the buzz and bluster around generative AI, one phrase has emerged as a rallying cry for the optimists and opportunists:

“English is the new programming language.”

As a metaphor, it’s clever. As a literal assertion, it’s misleading at best — and dangerous at worst.

Long before AI tools could turn prompts into Python or sketch UI in React, good programming started with one thing: clarity of thought. That often meant clearly expressed requirements, clearly articulated goals, and — yes — clear communication in English (or the primary language of the team).

So, in a sense, natural language has always been part of programming. But describing English as a replacement for actual code misunderstands what code does. Code isn’t just syntax — it’s structure, abstraction, and repeatability. It’s testable. It’s auditable. It’s modular.

AI may write code for you — but English doesn’t become code simply because you asked nicely.

Let’s be honest: AI can do a lot. It can refactor code, write documentation, scaffold web apps, even generate SQL queries. For low-risk, repetitive tasks — like CRUD scaffolding or boilerplate generation — it can be a genuine time-saver. But even then, someone needs to review the output.

There’s a critical catch: you’re still on the hook for the quality.

If you don’t understand the logic behind what it gave you — or if you can’t identify subtle mistakes, security flaws, or logic errors — then you’re trusting the output of a black box. That’s not programming. That’s gambling.

Ask yourself: If you can’t read the code it produces, how do you know it’s doing what you asked?

Would you trust AI to build an air traffic control system from an English prompt? Of course not. That kind of system involves real-time constraints, strict safety protocols, and intricate coordination between distributed systems. It demands precision, validation, and engineering discipline — not vague instruction.

Real applications live in ecosystems with business logic, performance constraints, regulatory standards, edge cases, and human lives in the balance. This is where the limits of AI become especially visible.

Take a records management system, for example — it isn’t just a collection of CRUD operations. It involves schema design, data retention policies, permission models, audit trails — the kind of decisions that aren’t “generated,” they’re designed.

And here’s the kicker: AI is excellent at sounding correct — but mediocre at being correct. That’s fine for brainstorming. It’s catastrophic for compliance.

Now, would you trust AI to help rewrite a patient-facing medical document in simpler terms? Maybe. With the right constraints, oversight, and review process in place, that task becomes more viable. The key difference is in the complexity and consequence.

Some tasks are genuinely enhanced by AI when paired with expert guidance. Others are too critical to be left to natural language prompting alone.

Ultimately, the claim that “English is the new programming language” conflates two very different things:

1. The convenience of instructing AI in natural language;
2. The rigor of software engineering, which involves far more than typing requests.

The former is a user interface. The latter is a discipline.

That’s not to diminish what AI can do — on the contrary, AI is one of the most powerful tools ever created for developers. It changes the way we approach certain kinds of work. But to say it replaces code? That’s a category error.

Ultimately, this isn’t just a misunderstanding of technology — it’s a strategic risk. Thinking that English can replace code encourages shortcuts, reduces accountability, and leads to brittle systems built on unchecked assumptions. The problem isn’t ambition — it’s oversimplification.

Let’s keep using English to communicate with AI — to explore ideas, generate drafts, and accelerate development. But let’s stop pretending it’s a substitute for programming expertise.

If we want to build systems that are reliable, secure, and scalable, we need people who understand how they actually work.

Because English is a tool — not the trade.