We’re All Panicking About AI. But What Even Is Programming?
I keep seeing the same panic pop up in developer Slack channels and Reddit threads. Someone posts a video of an AI generating an entire React app in 30 seconds, and suddenly everyone thinks software engineering is completely dead.
Honestly, we’re looking at this the wrong way. Programming isn’t dying. We just forgot what the word actually means.
Think about it. At its absolute core, programming has always just been finding a way to explain to a machine what you want it to do. That’s literally it. It’s just instruction delivery.
Back in the day, delivering those instructions meant physically punching holes in paper cards. You had to understand the hardware intimately just to do basic math. Then we got assembly, and later higher-level languages like C, Python, and Java. We invented object-oriented design and all these massive, headache-inducing architectures.
Why did we do all that? Because we were desperate for ways to give instructions faster and with less manual translation. Every single programming language we ever created was just another layer of abstraction to make talking to the computer less painful.
Right now, we are hitting the final layer of abstraction: natural language.
You don’t need to know the specific syntax of a REST API if you can just tell the system in plain English (or Russian, or Spanish) to grab the data you need. We are moving away from manually typing out the “how” and shifting entirely to focusing on the “what.”
This completely changes the game. Forget about building massive, static apps. If someone needs a tool to split utility bills with their roommates, they probably won’t download a dedicated app in the future. They’ll just tell their device, “Hey, figure out how much I owe Dave for the electricity and pay him.” The computer will spin up an autonomous agent, write the temporary logic, execute the task, and throw the code away.
So if software becomes infinitely cheap and easy to generate on the fly, what’s left for businesses? What’s the competitive advantage?
Compute.
If anyone can talk an AI into building a clone of a SaaS product in a single afternoon, your code is no longer a protective moat. The people who actually win this next decade are the ones sitting on the raw physical hardware. The massive data centers. The silicon chips. If you don’t have the compute to run these complex agent workflows at scale, you’re entirely at the mercy of the big tech companies that do.
Our jobs aren’t gone. They are just shifting. We’re going from syntax writers to agent managers. The new skill isn’t memorizing boilerplate code. It’s having the restraint and clarity to know exactly what to ask the machine to build, and verifying that it actually did it right.