Am I going to lose my job to AI?
The headlines cannot answer this for you, because the answer is not about AI in general. It is about the specific shape of your specific week — and that you can actually measure.
Most writing about AI and jobs is a horoscope: vague, alarming, and true of everyone and no one. It cannot tell you what you actually want to know, which is not "will AI get better" — it will — but "will it come for my job." That question has a real answer, and you can get to it without a crystal ball, because whether a job is automatable is mostly about its internal structure, not about how impressive the technology sounds this month.
The only question that matters
Here is the test, and it is simpler than the discourse around it. Divide your working week into two piles. In one pile: work that is executing a known procedure — following steps that could be written into a good instruction manual, applied to predictable inputs. In the other: work that is deciding under uncertainty and owning the result — figuring out what should be done when the manual runs out, handling the ambiguous case, being accountable when it goes wrong.
Now estimate the ratio. The larger your first pile, the more exposed you are, because the first pile is exactly what automation does well. The larger your second pile, the more durable you are, because judgment and accountability are the things a model does not hold. That is the whole assessment. It is uncomfortable precisely because it is honest.
If a manual could do your job, a machine soon can. If only judgment can, you are holding the durable half.
Reading your own result honestly
A few guardrails, because it is easy to flatter yourself here.
- Most jobs are a mix. Almost no one is purely one pile. The point is not a verdict but a direction: which pile is growing, and which way are you leaning?
- "It's complicated" is not the same as "it needs judgment." Plenty of complicated work is still procedural — long, fiddly, and fully specifiable. Complexity does not protect you. Genuine decisions under uncertainty do.
- The floor is rising regardless. Even durable jobs are losing their procedural bottom to automation. That is not a reason to despair; it is a reason to spend your effort on the half that is getting scarcer and better paid.
It is not a sentence. It is a map. Wherever your first pile is large, that is where to move your weight — toward deciding, building, and owning, and toward using the tools yourself so you direct the automation instead of racing it. The people who do fine are rarely the ones with the safest job today. They are the ones who read the direction early and moved.
What to do with the answer
If your honest ratio worried you, the move is not to hide from AI — it is to shift toward the work it cannot do, and to become the person operating it rather than the person it operates around. Concretely: take on ownership where you can, learn to build rather than only operate, and pick up a portable skill that rewards judgment.
In technology, that direction has a clear address. Designing, building, and running systems — rather than executing rote administration — is the durable side, which is why cloud engineering holds up better than routine IT work as automation spreads. If your test result has you thinking about a move, the ground is more reachable than the fear suggests, whether you are coming from support or changing fields entirely. The half that pays is the half you can still learn.
Common questions
How do I know if AI will replace my job?
Estimate what fraction of your week a good instruction manual could cover. If most of it could be handed to a new hire with that manual, that fraction is automatable. The part that is judgment, context, and accountability no manual captures is the durable part.
What jobs are safest from AI?
Work centered on judgment under uncertainty, accountability for outcomes, physical presence, or genuine human trust. In tech, that means designing and owning systems rather than executing rote tasks.
Is it too late to switch to something safer?
No. The durable, in-demand skills — building and automating systems on the cloud — are learnable around a full-time job, and they reward judgment over pedigree. Class One is free if you want to see whether cloud fits.
What should I do first if I'm worried?
Run the two-pile test honestly, then spend a few hours a week moving your weight toward the deciding-and-building side — including learning to use AI tools well yourself, so you direct them rather than compete with them.