Software Engineers are Freaking Out

quote:

---

Fifteen years ago, coding was supposed to be the future: As Marc Andreessen said that “software is eating the world,” nonprofits like Code.org popped up, raising $12 million from celebrities like Bill and Melinda Gates, and collaborating with the likes of Ashton Kutcher and Shakira, to teach underprivileged kids to program. In 2016, the Obama administration piloted a “Computer Science for All initiative, describing computer science as “a ‘new basic” skill necessary for economic opportunity and social mobility”; and, in the run-up to the 2020 election, Joe Biden went viral after he told a group of struggling miners that they should spend their days programming, rather than climbing inside a dirty hole....

And the people responded: Between 2009 and 2024, early-career computer science majors exploded from less than 70,000 to over 250,000; in the job market, the median base pay for developers grew by 24 percent between 2018 and 2024, until it was more than double the median salary in the U.S.

When Shawn started building software, it was because he thought—like most people—that it was a golden ticket. “I came from generational poverty,” he told me. His dad was a drug addict who disappeared for long stretches at a time, taking the rent money with him; his mother is disabled, and had a hard time finding work. When Shawn got into college, his biggest dream was financial stability, which is why he opted for computer science. That’s where the money is, he remembered thinking.

Now, the market has been flooded with AI agents like Anthropic’s Claude Code, OpenAI’s Codex, and Anysphere’s Cursor—all of which are capable of elite coding, and have wiped out the value of programmers’ skills. Today, the models are so good that anyone with an internet connection can be a “vibe coder”—just type in a prompt, and out comes a website, zero technical knowledge required.

Last November, when Amazon cut more than 14,000 jobs, software engineers in New York, California, New Jersey, and Washington were hit the hardest, accounting for nearly 40 percent of the layoffs across those states. In February, when Jack Dorsey’s company Block cut 40 percent of its workforce, he noted that “we've seen engineering work that would have taken weeks to complete be done by a small team in a fraction of the time.” Just last week, Oracle announced it too is laying off thousands, with software engineers significantly affected.

It’s gotten so bad that even the executives themselves are starting to mourn the loss of their craft: As OpenClaw founder Peter Steinberger recently put it, programming has become like knitting—something people do “because they like it, not because it makes any sense.” X’s Head of Product Nikita Bier posted, “We’re a few weeks away from where there will be no designers or engineers, but a third secret thing.

And the new technology was swallowing up all the entry-level positions for software engineers. Since ChatGPT’s public release, employment for software developers ages 22 to 25 has fallen by almost 20 percent; in 2025, the number of tech-job postings dropped by nearly one third compared to early 2020....

---

quote:

---

The fact is, if the AI takeover is already underway, computer science majors are best-placed to become experts in harnessing its power; there has never been a better time to learn the ropes of our brave new automated world than today.

For Daniel Bicalho, a 21-year-old computer science student, coding with a personal assistant is all he’s ever known. “My freshman year, AI could already do any coding assignment,” he told me. “It could knock out anything I got, by itself, literally in a minute or two.”

You’d think computer science professors would have noticed that the tools they’ve been teaching are no longer valuable. But Bicalho said they haven’t even budged. “The curriculum has still stayed the same,” he said. “I don’t think professors have found a way to work around AI yet.”

Now, with everyone on campus carrying around an automatic agent, and no one teaching them to harness it, Bicalho believes computer science students like him are complicit in de-skilling themselves. “You’ll find that a lot of these students don’t actually know how to code,” he said. “They’re just blindly writing AI slop.”

“There’s not a single employer that wants to hire them,” he explained, “because when it comes to discussing coding topics and fundamentals, they can’t actually say anything on the topic.”

Bicalho doesn’t reject the machine—in fact, as he told me, “I use AI for literally everything”—but he said he gets his competitive edge by acting as an editor: someone who knows that “their main purpose from now on is refining” their AI agent’s work—but knowing enough about coding to fix it if it goes wrong. After all, we need to preserve technical craftsmanship—the final masters of C++, fighting to keep human coding skills alive in case the machines suddenly blip out.

As Bicalho sees it, even if AI gets good enough to essentially carry out all the work we need doing without us intervening, the people in charge of it still need to know how it’s being done. Maybe the most worrying fact is that human beings are forgetting how the machines we invented work.

---