AI & human flourishing
The AI Anxiety Epidemic Is a Data Problem
Kenroy George · 2026-08-17 · 7 min read
TL;DR: AI anxiety persists because the public conversation offers only extremes, and nobody gives working people an actual number. Jobs are bundles of tasks, AI automates tasks rather than jobs, and your real exposure depends on your specific task mix. Once you can measure it, dread becomes a plan. Cari built a free 60-second test that does exactly that.
It usually arrives at 3am. Will AI take my job? Not jobs in general, not the labor market as an abstraction. Yours. The one that pays your rent, the one you spent fifteen years getting good at.
Almost everyone has asked it. Teachers, accountants, radiologists, truck drivers, copywriters, paralegals, software engineers who build the very systems they worry about. The question crosses every income level and every industry, and for most people it never gets a real answer.
The conversation is stuck at the extremes
Turn on the news or open a feed and you get two stories, told loudly and in turn.
Story one: nothing will really change. Every technology wave created more jobs than it destroyed, humans always adapt, relax.
Story two: everything will change. Entire professions will vanish within a few years, the disruption is total, panic.
Both stories are about everyone, which means neither is about you. And that is exactly why the anxiety never resolves. You cannot act on "everything" or "nothing." You can only act on your own situation, and nobody is measuring that.
This is the part worth saying plainly: the AI anxiety epidemic is not primarily a psychology problem. It is a data problem. People fear what they cannot measure.
Anxiety without measurement is dread
There is a real difference between fear and dread. Fear has an object. You can size it up, plan around it, act on it. Dread is fear without a shape, and it just sits there.
That is what the current moment produces. A vague, ambient sense that something large is coming for your livelihood, with no way to know how large, how soon, or how directly it points at you.
Measurement changes the category of the emotion. A number you can look at is something you can respond to. Anxiety without measurement is dread. Anxiety with measurement becomes a plan.
The same is true one level up. Journalists covering AI and work, and policymakers designing responses to it, are mostly working from the same two extreme narratives, because granular, person-level data barely exists in public. Better measurement would not just calm individuals. It would improve the entire conversation.
AI automates tasks, not jobs
Here is the key insight, and it is one that research on automation has pointed to consistently: a job is not one thing. It is a bundle of tasks.
A nurse assesses patients, documents care, calms frightened families, catches the anomaly the chart missed, and manages a dozen small human negotiations per shift. An accountant reconciles ledgers, but also advises a client through a bad year. A teacher grades papers, and also notices which kid stopped raising their hand.
AI does not arrive and take a job title. It arrives and takes tasks, one at a time, starting with the most routine and predictable ones. Which means two people with the same job title can face wildly different exposure, depending on how their working hours actually break down.
Your real exposure depends on three things. First, your task mix: how much of your week is routine information processing versus judgment, physical presence, and human trust. Second, your tools: someone already working with AI is positioned differently than someone whose workflow has not changed since 2015. Third, how you work: whether you are the person executing a process or the person accountable for its outcome.
None of that shows up in a headline about your industry. All of it shows up in a task-level measurement.
A number you can actually act on
This is why we built a free AI exposure test at Cari. It takes about 60 seconds. You tell it your occupation and answer a few questions about your actual task mix, and it scores your exposure from 0 to 100.
Two design decisions matter here.
The score is deterministic. It is computed from your occupation and task profile by a fixed scoring method, not improvised by a chatbot in the moment. Same inputs, same score, every time. If a number is going to inform decisions about your career, it should be reproducible, not vibes.
And the score is honest. If your exposure is high, the test says so. We think a comfortable lie is worse than an uncomfortable number, because the number is the thing you can act on. Nobody plans around reassurance.
What to do with a high score
Suppose you take the test and the news is not great. What then?
Not despair. A high exposure score does not mean your job disappears. It means the routine core of your job is automatable, and your move is to redirect toward the parts of the work that stay human.
Three categories reliably resist automation. Judgment: decisions made under ambiguity, where the data is incomplete and the stakes are real. Trust: relationships where a person needs another person to be accountable, to explain, to care. Exceptions: the messy cases that fall outside every process, which is where experienced humans have always earned their keep.
In practice that means shifting your effort, visibly and deliberately, toward those tasks. Volunteer for the ambiguous problems. Own the client relationships, not just the deliverables. Become the person who handles what the system cannot. And learn the tools, because in almost every field the near-term future is not a machine replacing a person. It is a person using machines replacing a person who refused to.
A low score is not a free pass either. It is a snapshot, and task mixes drift. The point of measuring is that you can measure again.
Common questions
Is a 60-second test really enough to measure this?
It is enough to replace a headline with a starting point. The test scores your occupation and task mix against a deterministic exposure model, which is a large step up from industry-level generalizations. It is a compass, not a prophecy, and it is free.
What if my score is high? Should I change careers?
Usually not, and certainly not as a first move. A high score means your current task mix is exposed, and task mixes can change without changing jobs. Redirect toward judgment, trust, and exceptions inside the role you already know, and adopt the tools before they are imposed on you.
Do I have to create an account or hand over personal data?
No. The test is free and takes about a minute. You answer questions about your occupation and how you work, and you get your score.
The future feels uncertain. Yours does not have to be.
The macro story of AI and work will stay contested for years, and the loudest voices will keep arguing about everyone in general and no one in particular. You do not have to wait for them to settle it.
The 3am question has an answer, or at least the beginning of one, and it is specific to you: your tasks, your tools, the way you actually spend your working hours. Measured, that anxiety turns into a list of next moves. Unmeasured, it just stays a weight.
If you want your number, take the free 60-second AI exposure test at ai.cari.global/future.