AI is not removing the first IT job. It is putting pressure on the scripted version of it. If your plan is to start in a role where you read a runbook, ask three qualification questions and escalate, that work is easier to automate. The first IT job worth training for is the one where you explain a strange result, connect two systems that do not normally talk, and make a reasonable call under uncertainty. Train for that one.
Who this is for
- Career changers choosing their first tech role and wondering whether AI has already closed the door.
- People in scripted first-line support who feel the seat getting tighter but are not sure what to move towards.
- Course sellers and bootcamps still marketing the old entry-level path.
What is actually changing
Parts of first-line support have been automated for years. Self-service portals, better knowledge bases and chatbots were already eating the easy cases. Large language models can now handle parts of the repeatable language work: drafting responses, summarising tickets, suggesting the next step from a runbook, rewriting a workaround in simpler terms.
This is a planning judgement based on what the tools can do, not a measured count of disappearing entry-level posts. Employers differ in adoption, reliability and staffing. But if you target scripted L1, ask how much of the job is learning to diagnose and how much is following a runbook.
The part that remains is the part the bot cannot do. The ticket that does not match any known pattern. The user who cannot explain the problem. The system where the logs say one thing and the user says another. The decision whether to reset a production account at 6pm on a Friday. These require judgement, context and ownership.
What people get wrong
The first mistake is confusing "AI can do part of my job" with "AI will take my job." AI can handle the repeatable part. It cannot handle the whole ticket. Across mainstream IT work, the boring slice compresses and the judgement slice expands. That may change the mix of work without telling us the net number of jobs.
The second mistake is the "learn AI to survive" panic. Most "AI engineer" roles require the systems background that career changers do not yet have. Skipping to the fashionable title without the foundational work is a worse bet than deepening the operational base. The durable AI-related paths are mostly underneath the headline: model serving, data engineering, GPU networking, MLOps. Those are not first jobs.
The third mistake is treating a cert as proof you can handle unfamiliar work. It can meet a named requirement, but evidence of how you handle ambiguity makes the story stronger.
What to do instead
Pick a first role that puts you next to uncertainty. A small internal IT team where you touch everything. A managed service provider with varied clients. A T2 support desk that sees the tickets the T1 bot could not close. A field or desktop role with physical access and real users. The common thread is that you see situations the knowledge base does not cover.
Build two skills in parallel alongside any cert. The first is diagnostic writing: can you describe a fault, what you have tried and what you still do not know in three sentences? The second is tool fluency without tool worship: use AI to draft responses and explain concepts, but never let it make the decision you will be asked to defend.
Treat the first role as an apprenticeship and review what you are learning as you go. Plan a possible sideways move: into infrastructure, security, cloud support, networking or identity. The risk is staying with only scripted work and no exposure to harder cases.
When to walk away from the plan
If your only realistic local option is pure scripted L1 and you cannot relocate, take it. But use it as a foothold and look for mentoring and harder tickets. If a course is selling you a "junior AI engineer" outcome without a coding or systems background, that is a different kind of shrinking seat.
Where this connects on TechWaymark
For the broader argument on why AI shrinks specific IT roles rather than all of them, read AI will not delete IT. For why IT support is still the right starting point for most people, read Why most people should start in IT support. The pathways page tags each route with editorial AI-exposure guidance based on the work itself, not a jobs forecast.