The history of computer programming is a long journey in abstraction. 50 years ago, writing a program was punching holes in a stack of cards, handing them to an operator, and heading home while the computer ran all night. If there was a bug, you'd make the changes and go through the whole thing again. Then came assembly, then high-level languages like COBOL, then visual programming, and on and on until you see a kid in a Brooklyn coffee shop shouting into his phone at someone named Claude.
As time went on and these new advancements appeared, inevitably there would be a moment where someone would proclaim that developers were finally being automated away. No more heavy sighs from the bullpen of keyboard jockeys. The nerds have lost… huzzah! Except they hadn’t and they continued to build.
We now find ourselves in another one of those moments. In December of 2024, Mark Zuckerberg went on Joe Rogan and said that AI at Meta was already reaching the capabilities of mid-level software engineers. By 2025, he said, most of the code in Meta's apps would be written by AI instead of people.
A tech leader making bold claims is one thing, but the wave of people gleefully "dunking" on the developer during the rise of AI has been… enlightening. Developers were not only justifiably worried about layoffs, they were disturbed by the manic enthusiasm for them.
Still, this is not going to be a rage post about tech villains. A lot of the people saying naive and lazy things about AI were just following the hype cycle and repeating wild claims from people excited about the shiniest new tech. This is what drives LinkedIn after all. Still, it was quite a moment to live through for a developer. For about 18 months (and counting), if you were a working developer, you couldn't open a tab without someone in the C-suite explaining how AI would not only soon cure cancer, it would also make the builders of their products unnecessary. You probably have your own examples in mind.
Indeed, the hype cycle had consequences. Companies didn't just talk about replacing developers, many actually did it. And for many, things played out differently than they thought they would.
IBM is a good example of the lifecycle of the AI hype. In May of 2023, CEO Arvind Krishna told Bloomberg that AI was going to let the company eliminate 7,800 back-office jobs over the next five years. It was the first major Fortune 100 announcement of its kind and it set the tone for everything that followed.
Smash cut to two years later, when Krishna told the Wall Street Journal that AI had actually touched a "couple hundred" HR workers, not 7,800. IBM's total headcount, he said, had gone up, not down. The savings got reinvested in software engineering, marketing, and sales. By February of 2026, IBM's CHRO was announcing the company would be tripling entry-level hiring "for software developers and all these jobs we're being told AI can do." To be fair, I give credit to IBM for course-correcting, even if they were part of the initial panic.
And it wasn't just IBM. More than a third of employers in a recent Forrester study spent more on restaffing than they pocketed from the original cuts. In the rabid rush to gut development teams, it was clear they'd bought into a half-baked story based more on dreams than reality. The fact that the dream was to eliminate developers didn't go unnoticed.
Whatever the actual motive, the effect on developers was the same. Some of these tech leaders genuinely believed AI was going to replace their engineers. Others were using the AI story as cover for layoffs during an erratic economy. Either way, a lot of developers spent 2024 and 2025 being told their jobs were ending by people who didn't seem particularly sorry about it.
So what's the practical view, then?
If it's not obvious already, this isn't an anti-AI piece. I use Claude Code every day and Claude Co-work to help my ADHD-addled brain keep up to speed on what I need to do. The company I work for ships AI-accelerated developer tooling. Much of the AI hype was valid but misdirected and clumsy. Used properly, AI is a multiplier for developers who already know what they're doing, whether they're full speed ahead and no longer touch code, or they prefer writing their own code but use AI to diagnose, debug, and strategize. Why simply cut developers for a cash grab when you could use them to ship features and improve code at an accelerated rate? Are we on defense or offense?
If anything, that work is getting harder to do without experience. The 2025 DORA report found that AI amplifies the quality of work. Good engineering teams will continue to ship good code, but at a faster rate. Bad engineering teams will continue to ship bad code, but also at a faster rate, piling up technical debt faster than ever before.
So what does it mean for a company to be pro-developer in the age of AI? It's a set of choices that show up in how a company actually operates.
It means respecting the leverage. A senior developer who can now work more efficiently is worth more than they were before, not less. The math of replacing them with cheaper labor and an AI subscription was wrong from the start, and those that know this are keeping their good engineers happy and hiring more of them.
It means making AI fit into the developer's workflow, not the other way around. Early in the hype cycle, the people saying "AI is a great tool" got drowned out by the ravenous enthusiasm of "AI can replace everyone." Voices of reason aren't as loud but are often right. The smart path is asking the people doing the actual work which AI features help them ship and which ones get in the way, and then building accordingly.
It means being honest about whether AI is a toy, a tool, or a replacement. There are real applications of all three, and pretending otherwise is what got us into this sloppy hot take hot mess. Some workflows really are better off automated end-to-end. Some really do need a human at every step. Most are somewhere in between, with an AI doing the boilerplate and a human owning the outcome. Figuring out which is which is an engineering question. It deserves more nuance than a clickbait quote from someone looking to talk up their valuation on a podcast.
To be clear, this isn't cut and dry. AI is a complicated topic with real concerns I haven't touched on here, from regulation to energy use to jobs that really are becoming less viable. There is a wide spectrum of sentiment amongst developers on AI. Some are all in, some hate it with the fire of a thousand suns. Most are somewhere in the middle, using the tools every day and trying to figure out what the next few years actually look like.
The fact of the matter is that the carnival barkers had their moment, and the moment is ending. What's left is the work, and the people doing the work to build software that does what it's supposed to do, for customers who are counting on it, at companies that need to keep shipping. The tools change, the abstractions get higher, and the people who actually understand what they're building keep mattering. Maybe some people at the top can learn from this and remember that what seems disposable today might be essential tomorrow. Either way, there are people underneath it all.
Let's just get to work and whatnot.