AI Has Learned to Code and Is Taking Over

A month ago an entrepreneur named Matt Schumer wrote a lengthy post on X titled “Something Big is Happening.” I wrote about it here. The gist of it was that AI tools had recently made a quantum leap, especially when it came to writing computer code. Schumer claimed that he no longer needed to do the work, he just need to tell the AI what he wanted done and it could do it for him.

I am no longer needed for the actual technical work of my job. I describe what I want built, in plain English, and it just… appears. Not a rough draft I need to fix. The finished thing. I tell the AI what I want, walk away from my computer for four hours, and come back to find the work done. Done well, done better than I would have done it myself, with no corrections needed. A couple of months ago, I was going back and forth with the AI, guiding it, making edits. Now I just describe the outcome and leave.

Because the post went viral, there was a lot of discussion about whether Schumer was overhyping it or if AI really had crossed some sort of threshold. Since then I’ve at least a couple more article basically agreeing that Schumer is right. Coding seems to be one of the things AI is best at and that’s completely changing how code gets written. The NY Times has a new story about it today which reads like a confirmation of most of what Schumer claimed.

Lately, Manu Ebert has been trying to keep his A.I. from humiliating him.

I recently visited Ebert, a machine-learning engineer and former neuroscientist, at the spare apartment where he and Conor Brennan-Burke run their start-up, Hyperspell. Ebert, a tall and short-bearded 39-year-old with the air of a European academic, sat before a mammoth curved monitor. Onscreen, Claude Code — the A.I. tool from Anthropic — was busy at work. One of its agents was writing a new feature and another was testing it; a third supervised everything, like a virtual taskmaster. After a few minutes, Claude flashed: “Implementation complete!”…

Ebert, initially cautious, began letting it do more and more. Now Claude Code does the bulk of it. The agents are so fast — and generally so accurate — that when a customer recently needed Hyperspell to write some new code, it took only half an hour. In the before times? “That alone would have taken me a day,” he said.

The AI doesn’t always get it right but then neither do human coders. Ebert finds that he just needs to make it really clear to the AI that it’s not allowed to skip any steps and part of that means telling it not to embarrass him.

I looked at Ebert’s prompt file. It included a prompt telling the agents that any new code had to pass every single test before it got pushed into Hyperspell’s real-world product. One such test for Python code, called a pytest, had its own specific prompt that caught my eye: “Pushing code that fails pytest is unacceptable and embarrassing.”

Embarrassing? Did that actually help, I wondered, telling the A.I. not to “embarrass” you? Ebert grinned sheepishly. He couldn’t prove it, but prompts like that seem to have slightly improved Claude’s performance.

The irony of coders being the first jobs that AI can really replace is noteworthy but the tradeoff for the guys who already work in the field is that they can get a lot more done in less time than they every could before.

“We’re talking 10 to 20 — to even 100 — times as productive as I’ve ever been in my career,” Steve Yegge, a veteran coder who built his own tool for running swarms of coding agents, told me. “It’s like we’ve been walking our whole lives,” he says, but now they have been given a ride, “and it’s fast as [expletive].”

But in going that fast the nature of the work has changed to. It’s no longer about writing in Python. Now you just tell Claude what you want the code to do and it does all the grunt work.

A coder is now more like an architect than a construction worker. Developers using A.I. focus on the overall shape of the software, how its features and facets work together. Because the agents can produce functioning code so quickly, their human overseers can experiment, trying things out to see what works and discarding what doesn’t. Several programmers told me they felt a bit like Steve Jobs, who famously had his staffers churn out prototypes so he could handle lots of them and settle on what felt right.

That’s really just the start of this lengthy article which is focused on small start ups and individual coders. Those folks really are zooming along faster than was ever possible before. But at big companies like Google, the goal isn’t to revolutionize the code every few months, it’s to keep the millions of lines of code you already have working as you gradually add features. And it turns out AI is good at that too.

The whole thing is worth reading but the bottom line is that we really have turned a corner here. There is now one white collar job that AI can do as well or in some ways (certainly in terms of speed) better than a human, and that’s coding. We should start seeing the acceleration that comes from that as the year progresses.

Editor’s Note: Do you enjoy Hot Air’s conservative reporting that takes on the radical left and woke media? Support our work so that we can continue to bring you the truth.

Join Hot Air VIP and use promo code FIGHT to receive 60% off your membership.