Three groups of engineers screaming themselves out of a job

Three groups of engineers are about to lose their jobs and they won’t shut up about it.

I don’t mean this as a prediction. I mean it as an observation about a pattern I’ve watched repeat across companies, teams, and conversations for the last two years. The pattern is: engineers who feel most threatened by AI are also the ones most aggressively broadcasting that threat. And in doing so, they’re demonstrating exactly the kind of thinking that makes them easy to replace.

Here are the three groups.

Group one: the company that doesn’t know what to do with AI

You’ve seen this place. Management sends an email about “embracing AI” that’s vague enough to mean nothing. A few engineers experiment. Most don’t. The sprints don’t change. The processes don’t change. Velocity doesn’t change. The AI tools are there — they sit in a corner, half-used.

The engineers here aren’t in danger of losing their jobs because of AI. They’re in danger of losing them because the company is slowly becoming uncompetitive, and they’re not pushing back. The complaint from engineers in this environment is usually: “They’re not taking AI seriously enough.” Which is true. But the solution isn’t to shout about it in Slack or on LinkedIn. It’s to do the thing yourself, build something with it, and make the output visible. Every engineer who waits for organizational permission to use AI effectively is already a year behind.

Group two: the company that refuses

This one is simpler and rarer but real. The company has decided, explicitly or implicitly, that AI-assisted development isn’t something they want. Sometimes this is about IP. Sometimes it’s about culture. Sometimes it’s about a CTO who’s made a bet and doesn’t want to be proven wrong.

The engineers here feel this as a trap: they can see that the outside world is moving faster, they can see that their skills are getting narrower relative to what the market wants, and they can’t do anything about it in their current context. The noise from this group sounds like: “I can’t believe we’re not allowed to use these tools.” Which is a reasonable frustration. But the action it implies — staying and complaining — is the wrong one. The correct action is to leave as soon as you have somewhere better to go, build the skills externally, and treat the constraint as temporary.

Neither of these groups is the loudest. Group three is the loudest.

Group three: the manager already figured it out

This is the group that concerns me most, because they’re the most sophisticated and the most confused at the same time.

These are senior engineers, often with strong opinions about craft, who have correctly identified that AI is changing what “good engineering” looks like — and have concluded that the right response is to argue loudly for the value of human expertise. You see them in conference talks, long-form posts, and Twitter threads explaining why AI can’t really understand context, why the code it produces is unmaintainable, why the real skill is knowing what to build, not how to build it.

All of those points have merit. None of them are the point.

The manager already figured it out. The product leader already figured it out. The startup founder already figured it out. What they figured out is not “AI writes perfect code” — it’s “AI changes the unit economics of software delivery in a way that changes how I staff, what I expect, and who I hire.” That’s the actual shift. And engineers who are publicly demonstrating that they need a lengthy argument to accept that shift are, in the process of making that argument, demonstrating exactly the brittleness that makes them the obvious cost-cut.

What all three share

The common thread across all three groups isn’t fear, exactly. It’s a confusion between what they’re good at and what they’re being paid to do.

Being good at writing code is not the same thing as being paid to write code. You’re paid to ship things, to make decisions, to hold quality bars, to reduce the cost and risk of delivery. AI changes how those things get done. It doesn’t change that they need to get done.

The engineers I’ve watched not have this problem share one characteristic: they stopped thinking about AI as a threat to their craft and started thinking about it as a change to their job. That’s a smaller and more tractable problem. You adapt your job all the time. You just usually don’t call it existential.

The screaming is the tell. Confident people don’t scream about threats. They make moves.