Last week I wrote about a viral essay titled "Something Big is Happening." The essay was written by an entrepreneur named Matt Schumer and he later explained that he'd written it for his father, an AI skeptic, to try to explain to him how fast the technology was moving. What he described was a recently leap in AI's abilities, especially when it came to coding:
...on February 5th, two major AI labs released new models on the same day: GPT-5.3 Codex from OpenAI, and Opus 4.6 from Anthropic (the makers of Claude, one of the main competitors to ChatGPT). And something clicked. Not like a light switch... more like the moment you realize the water has been rising around you and is now at your chest.
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.
Today the NY Times published an opinion piece by Paul Ford which reads like it's meant to be a confirmation of what Matt Schumer wrote last week.
On weekday evenings, heading home on the subway from Union Square in New York City, I log into an A.I. tool from my phone, and write a prompt. “Look at the data in the files I just uploaded,” I tap. “Load it into a database, then make it searchable with a web interface.” Underground in the subway tunnels my internet connection drops, but when my train emerges onto the Manhattan Bridge, I get a few minutes to see all the work my coding agent has done, and if I type fast enough I can issue another prompt. By the time I get home to Brooklyn, my little project tends to be done: A website, a feature in a music app, a complex search tool or some tiny game.
This is called “vibe coding,” a term coined a year ago by the artificial intelligence expert Andrej Karpathy. To vibe code is to make software with prompts sent to a specialized chatbot — not coding, but telling — and letting the bot work out the bugs. Like many other programmers, I use a product called Claude Code from Anthropic, although Codex from OpenAI does about as well, and Google Gemini is not far behind. Claude Code earned $1 billion for Anthropic in its first six months. It was always a helpful coding assistant, but in November it suddenly got much better, and ever since I’ve been knocking off side projects that had sat in folders for a decade or longer. It’s fun to see old ideas come to life, so I keep a steady flow. Maybe it adds up to a half-hour a day of my time, and an hour of Claude’s.
November was, for me and many others in tech, a great surprise. Before, A.I. coding tools were often useful, but halting and clumsy. Now, the bot can run for a full hour and make whole, designed websites and apps that may be flawed, but credible.
Again, this is very close to what the viral article claimed was happening. The main difference is that Schumer sees the big jump happening a few weeks ago and Ford thinks it happened last November. Either way, the bottom line is that these tools can now create mostly finished projects with no coding at all, just a few written directions about what to do. And both authors agree that they're seeing the AI capable of working for a significant amount of time without additional input from a human. Today that's only an hour or a few hours but as Schumer noted last week, that metric is doubling every few months.
There's an organization called METR that actually measures this with data. They track the length of real-world tasks (measured by how long they take a human expert) that a model can complete successfully end-to-end without human help. About a year ago, the answer was roughly ten minutes. Then it was an hour. Then several hours. The most recent measurement (Claude Opus 4.5, from November) showed the AI completing tasks that take a human expert nearly five hours. And that number is doubling approximately every seven months, with recent data suggesting it may be accelerating to as fast as every four months.
Ford breaks down the amount of money that a software company might charge for the kind of work that AI can now do with a few prompts.
When I rebooted my messy personal website a few weeks ago, I realized: I would have paid $25,000 for someone else to do this. When a friend asked me to convert a large, thorny data set, I downloaded it, cleaned it up and made it pretty and easy to explore. In the past I would have charged $350,000.
That last price is full 2021 retail — it implies a product manager, a designer, two engineers (one senior) and four to six months of design, coding and testing. Plus maintenance. Bespoke software is joltingly expensive. Today, though, when the stars align and my prompts work out, I can do hundreds of thousands of dollars worth of work for fun (fun for me) over weekends and evenings, for the price of the Claude $200-a-month plan...
And for lots of users, that’s going to be fine. People don’t judge A.I. code the same way they judge slop articles or glazed videos. They’re not looking for the human connection of art. They’re looking to achieve a goal. Code just has to work.
Do we want "vibe code" running the Pentagon? Absolutely not. But is it good enough that a lot of lower level commercial work could be done a lot cheaper than it could two years ago? Probably so. And that is going to translate into lost jobs for some people who were making a killing designing website and who now won't be able to make a fraction of that amount because AI is driving the price down.
And the AI models get better fast. It may be coding this month but by the end of the year it could be law or some other specialty where the AI makes a sudden leap.
No matter where you work, my hunch is this is coming for you. Have you noticed the software you use every day adding “A.I. features”? That’s the top of the slippery slope. Whatever unifying principle equates to ship risk in your industry, people are trying to mitigate it with A.I. Insurance, finance, architecture, manufacturing, textiles, every kind of project management — they want to automate it all through A.I...
The simple truth is that I am less valuable than I used to be. It stings to be made obsolete, but it’s fun to code on the train, too. And if this technology keeps improving, then everyone who tells me how hard it is to make a report, place an order, upgrade an app or update a record — they could get the software they deserve, too. That might be a good trade, long term.
Good for the economy but bad for a bunch of individual workers who've spent years making a living as coders. That's going to be the pattern moving forward.
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.
Join the conversation as a VIP Member