Will AI replace developers, testers, and support engineers?
AI won't replace these roles wholesale — but it is quietly deleting the repeatable half of each one. Here's what survives in development, testing, and support, and how to move toward it before the compression reaches you.
Key takeaways
- AI removes tasks, not roles — but if your role is mostly the removable tasks, that's a distinction without a difference
- In development, the survivor skill is judgment and ownership
- in testing, it's risk thinking
- in support, it's handling the hard, ambiguous cases
- The safest position in every one of these roles is the part that requires understanding *why*, not just *how*
- Move toward the surviving half deliberately — waiting for the compression to reach you is the expensive option
Every week an experienced engineer asks me some version of the same question: "Is my job next?" The honest answer is that AI is unlikely to replace your role outright — but it is already deleting the repeatable half of it, and in a lot of teams that half was most of the day. Whether that feels like "AI replaced me" depends entirely on where you were standing.
Let's take the three roles people worry about most.
Developers
What AI does well: boilerplate, common patterns, first drafts, translating between languages, explaining unfamiliar code, and generating obvious tests. For a developer whose day was mostly "implement this well-specified ticket," that's a large chunk of the work, done faster.
What AI does poorly: deciding what not to build, untangling a vague requirement into something buildable, judging whether a generated solution is correct in this system, owning a production incident at 2 a.m., and making the architectural call that will still make sense in two years. The survivor skill in development is judgment and ownership — being the person who directs and verifies the work, not the person who out-types the tool.
Testers
AI can generate test cases, produce test data, and draft automation scripts quickly. If your value was writing large volumes of straightforward test cases, that's the part under pressure.
What it can't do is think about risk: what's most likely to break, what failure would hurt the business most, which edge case nobody wrote a requirement for, and whether "all tests pass" actually means "the product is safe to ship." The survivor skill in testing is risk thinking — quality strategy over quality execution. Testers who understand the product and the user, and who can say "here's what I'd worry about," become more valuable when everyone can generate cases cheaply.
Support engineers
Retrieval assistants and AI agents now handle a growing share of L1: password resets, known issues, documented fixes, first-response triage. That layer is compressing fastest of the three.
But the hard cases — the angry enterprise customer, the novel bug the knowledge base has never seen, the issue that's really three issues tangled together, the judgment call about escalation — still need a human who understands the system and the customer. The survivor skill in support is handling ambiguity and the difficult human moment, plus the ability to turn recurring pain into a permanent fix rather than answering the same ticket forever.
The pattern across all three
Notice the common thread. In every role, AI takes the part that follows a known procedure, and leaves the part that requires understanding why — judgment, risk, ownership, ambiguity. That's not a coincidence; it's the shape of the whole shift. The repeatable layer compresses; the judgment layer becomes scarce and better paid.
So the question isn't really "will AI replace me?" It's "how much of my current role lives on the repeatable layer, and how fast am I moving off it?" If you're honest that the answer is "most of it, and slowly," that's not a reason to panic — it's a reason to make a deliberate plan.
Moving before the compression reaches you
The professionals who navigate this well don't wait. They map where they want to sit — deeper technical specialisation, a move into management, or a process/product role — and build toward it on purpose. The ones who struggle are usually reacting after a layoff, buying certificates in a hurry with no direction. The ProThink Experienced Professionals workshop exists precisely to build that plan before you need it.
FAQ
Are AI tools making my IT role disposable?
Not the role — the repeatable part of it. AI compresses well-defined, procedural tasks in development, testing, and support. The judgment, ownership, and ambiguity-handling parts of each role become more valuable, not less. Your exposure depends on how much of your day is procedure versus judgment, and how fast you're shifting toward the latter.
Why are competent mid-level IT professionals still getting laid off?
Because "good at the task" and "positioned where the value is" are different things. If your role sits mostly on the repeatable layer that AI now handles, being excellent at that layer isn't protection — it's the exact work being compressed. Resilience comes from where you stand on the value chain, not effort.
Should I reskill or switch companies to stay safe?
Neither by itself. Switching companies with the same profile just moves the same exposure elsewhere, and reskilling without direction wastes money. Decide where you want to sit on the value chain first, then reskill toward that specific position. A plan makes both moves effective instead of hopeful.
How do I know if I'm at risk of being laid off in IT?
Ask what share of your week is repeatable, procedural work that an AI tool could now do, versus judgment, ownership, and ambiguous problem-solving. A high repeatable share signals exposure regardless of how well you do it. The fix is deliberately moving toward the judgment layer, not working harder on the automated one.