Will AI take my job? Four takes this week and not one of them named your actual Tuesday. That question gets settled here, with a test you can finish before your coffee cools.
Start with the marketing coordinator at $62,000. The title sounds like taste and relationships, but roughly 55% of the week is drafting copy and reports against a template. Score it: pattern work 1, a manager signs off before publishing 2, weekly client meetings 1. That is 4 of 6, the high band, and nothing about the work got less skilled on the way there.
Three prompts, scored 0 to 2 each: how much of the week is pattern work, who is accountable for the output before it leaves, and how much of the job needs you physically or relationally present. One page, under two minutes, no account, no email.
You walk away with a total from 0 to 6, a band, and one move sized to a month. A number written down instead of a mood. Exposure belongs to your week, not your title, and three countable things put a figure on it.
Quick Tips Before You Start
Use last week's calendar
Pull up the actual week you just worked, not a typical week you remember.
Set a two-minute timer
Score fast. Overthinking turns evidence into a guess.
Skip the job title
Your business card is a claim; your sent folder is the evidence.
Score each factor 0-2
Three numbers, total 0-6. No half points.
Write the date
A score is a snapshot. Date it so you can compare in 90 days.
1. Your job title is not what gets automated
Tasks get automated, not titles. That is the whole premise of the test, and it is why the first question never asks what you do for a living.
The score looks at exactly three things: the share of your week spent on pattern work, whether someone accountable signs off before the output goes out, and how much of the job needs a person physically or relationally present. Occupation, industry, years in, none of it is asked. Three prompts, each scored 0 to 2, added to a total of 0 to 6, on one page, finished in under two minutes, with no account, no email and no login to create before the first question.
A content designer is the clean example of a title lying in the direction of exposure. On a business card, the role reads as taste: voice, tone, brand judgment. On the timesheet it is a week of filling a fixed social template and a product-blurb template in a shared doc, slot after slot, and the sent folder knows it.
The error runs the other way just as often. An engineer or systems analyst whose week is site visits, contractor walkthroughs and signing off on other people's drawings sounds like the first thing a model eats. It is actually the two safest things on the list: a signature and a room.
The business card is a claim about the job. The calendar and the sent folder are the evidence, and only the evidence gets scored.
💡 Tip: Write down the three factors on a sticky note and keep it visible while you score. It stops you drifting into title guesswork.
| Score | What it means | First move |
|---|---|---|
| 0-1 (low) | Little pattern work, little sign-off, little presence needed | Log tasks for two weeks in half-hour slots |
| 2-3 (moderate) | Some template-shaped work under a light review | Add a review or strategy step to the most pattern-heavy task |
| 4-5 (high) | Heavy pattern work with a sign-off gate and some client contact | Shift one task to oversight or client-facing work within 30 days |
| 6 (very high) | Most of the week is template work with no human gate or room | Test one task from an adjacent role this month |
2. How much of your week is pattern work?
Count the hours you spend producing text, images, code or numbers against a recognizable template. That count is the number, and a competent model already knows the shape of the answer.
The work has a specific feel. Drafts, summaries, standard monthly reports, routine code, templated designs: anything where the second version looks like the first with new inputs. A finished doc that took you three hours but followed last month's structure is pattern work, and it is what the score is measuring.
The anchors are blunt. Under 20% of the week on that work scores 0. Between 20% and 60% scores 1. Over 60% scores 2.
The marketing coordinator at $62,000 spends about 55% of the week drafting copy and reports against templates. That lands in the middle band: a 1.
Volume beats skill, and this is the catch most people miss. A fast, excellent writer of routine copy has more template-shaped output to hand over, not less. The model's advantage is measured against the amount of standard work produced, so the quickest producer walks in with the most of it.
This number sets the ceiling on how much of the job a model can draft. Whether anyone lets it is a separate question, answered by the two that follow.
⚠️ COMMON MISTAKE
Don't count meetings as pattern work
A weekly status call that a calendar invite could replace is not pattern work and it is not presence. Score the hours you actually produce something against a template. The rest belongs to question three.
3. Who checks your work before it matters?
A draft can sit in your folder forever, but whether it ever leaves depends on whose name has to be attached when it does.
Score the gate on three levels. If nothing at all rides on the output, that scores 0. If a colleague glances at it before it goes out, that scores 1. If someone accountable signs off before it reaches a client, a regulator or a customer, that scores 2. The coordinator's manager signs before anything publishes, so that is a clean 2.
Liability is the brake, and it is stubborn. A contract, a set of medical notes, a safety call, a regulated filing: someone has to stand behind the signature, and a model brings no licence, no insurer and no one to be held responsible. Drafting inside those gates stays human longest, not out of sentiment but because there is nowhere to file the blame.
Sign-off that is pure rubber-stamping is not protection, though. A reviewer who only skims gets replaced before the writer does, because the skimming is the part a tool can mimic. The gate secures the person who owns the judgment, which means a 2 tells you accountability sits somewhere. The question to ask is whether it sits on you.
The useful fact in this number is not how strict the review is. It is whose name is on it. Drafting that someone else signs is the substitutable part of your job.
💡 Tip: If your name is on the sign-off, say so out loud in your next review. That is the part of the workflow a model cannot absorb.
What the score measures vs. what it doesn't
It measures
- How much of your week is template-shaped output
- Whether an accountable person signs off before it ships
- How much of the job needs you physically or relationally present
It does not measure
- How fast your employer will adopt AI tools
- What happens to pay or headcount at your company
- How quickly you can learn the adjacent tool
- Your ability or your worth as a worker
4. How much of the job needs a person there?
Sign-off decides whether a draft passes the gate, but a sizeable slice of the working week never turns into a document in the first place, and that slice is what this last number captures.
Scoring here runs 0 to 2. No human contact anywhere in the role scores 0. Occasional client calls, walkthroughs or relationship hours score 1. A job where most of the hours involve bodies on a site, trust held in a name, or people in a room scores 2.
Take the marketing coordinator at $62,000. Her client meetings run about once a week, so she scores 1 here, and a 1 is the honest answer for a role where most of the week is screen time.
Relationship work counts even through a laptop camera. A client who calls and asks for her by name rather than phoning the company line is proof of trust stored in a human, and a chatbot answers the line without a face to place.
This number gets inflated more than the other two, so score it carefully. A Tuesday status call that a recurring calendar invite and a 10-line summary could stand in for is not presence, and a meeting-heavy calendar is not evidence of a durable job except where those meetings would genuinely disappear if a screen took the seat.
Get it wrong in the other direction and the cost is real: new hires who overclaim presence, underprepare the honest number, and then mis-size their next move by a full band or more.
For most readers, presence is the number that offsets heavy pattern work, which is exactly why it is the one worth scoring twice before trusting it.
Score each factor 0-2. Total 0-6. No half points.
5. Add up the score and read the band
Add the three numbers and the total, not any one factor, is what names the band.
The mapping is flat and memorable: 0 to 1 lands low, 2 to 3 moderate, 4 or 5 high, and a clean 6 means very high. Three mid-to-high scores produce a high total even where one factor looks manageable on its own.
Walk the coordinator through it. Pattern work 1, since 55% of her week is drafting; sign-off 2, since her manager approves before publishing; presence 1, from the weekly client calls. That gives 4 of 6, which is the high band.
That 4 is not a forecast of a layoff, not an employer's decision, and not a judgment on skill. It just tallies how much of the week lives in task shapes that can be passed to a tool.
Two people can both score 4 and still face wildly different timelines. One employer has already pushed an AI tool into the workflow and set output quotas against it, while the other is still running the 2019 process with the same sign-off chain, and AI's impact on jobs at a given company is a fact about that employer, not about the score.
Read the band as a label for what to do next. It is not a diagnosis of what will happen, and no number moves on its own without an action attached.
📅 Your 30-day move by band
Low (0-1): log tasks
For two weeks, record every task in half-hour slots so your next score is evidence.
Moderate (2-3): add a review step
Attach a review, editing, or strategy step to your most pattern-heavy task.
High (4-5): shift one task
Within 30 days, move one task to oversight or client-facing work and tell your manager.
Very high (6): test one adjacent task
This month, try a single task from a neighboring role before deciding anything bigger.
6. What to do first at each band
Each band comes with one action sized to a single month, because a label is only useful if it points at a Monday task.
Low, 0-1: log your tasks for two weeks in half-hour slots, recording the actual work rather than the job title, so the next score rests on evidence instead of memory. A banker who writes 'client servicing' on the self-review but spends 4 hours a day in a recurring reporting template will see the real number once the log fills in.
Moderate, 2-3: bolt a review, editing or strategy step onto the most pattern-heavy task you do. If the weekly newsletter goes out from a fixed structure, try Perplexity AI prompts to speed the draft, then take on the performance read that follows it, so your name sits on the judgment rather than the draft.
High, 4-5: within 30 days, move one task to oversight or client-facing work and tell your manager in plain language. The coordinator takes the campaign performance review and hands off the monthly report, and the swap shows up on her calendar before the quarter ends.
Very high, 6: pick one task from a neighboring role and do it once this month before committing to anything larger. One piece of the analytics handoff is enough to test the water, and it is not a resignation letter and not a paid bootcamp.
This advice skips two groups. Someone inside a year of retiring does not need a 30-day repositioning exercise, and a licensed role such as nursing, law or electrical work has its own rules, its own board, and a timeline set by the licence and the employer's process, so a single month is the wrong unit to plan around.
💡 Tip: Say the move out loud to your manager. A spoken commitment is harder to quietly drop than a private plan.
💡 PRO TIP
Date the score and keep it
Write the date next to your total and revisit it in 90 days. A score is a snapshot of your week as currently spent, and weeks get reassigned quietly. The dated note is how you notice.
7. What the test cannot tell you
The test leaves three things unmeasured: how fast your employer will actually roll the tools out, what happens to pay and headcount at your specific company, and how quickly you can pick up the adjacent tool.
The first blind spot is adoption speed, and it is a property of the building you work in, not of your week. One employer has already put an AI assistant in every workflow and set productivity targets against it; the other is still running the 2019 process with the same weekly template. Same score, different clocks.
The second is money. A 0-6 number says nothing about whether your company freezes a headcount, trims a contractor line, or quietly reassigns the drafting to whoever already sits nearest the tool. Those calls happen in budget meetings the score never sees.
The third is you. How fast you can learn the adjacent tool – the analytics dashboard, the review workflow, the client-facing deck – is a personal clock the test cannot read.
For the parts a snapshot can't reach, go to the source. The US Bureau of Labor Statistics Occupational Outlook Handbook and O*NET carry occupation-level outlook and task content, and that is where to check what the role trends toward rather than guessing it from a total of 4.
A low score is not a green light. Under 20% pattern work means the week currently sits light on template output, and the score is a snapshot of that week as spent – reassignments happen quietly, without a title change or a meeting about it.
So spend the two minutes now, write the number on the page, and make the move your band calls for inside 30 days. The test is a starting line, not a finish line.
Frequently Asked Questions
How long does the test actually take?
Under two minutes. Three prompts, each scored 0-2, on one page. No account, no email, no login to create before the first question.
What if I land between two scores?
Use whole numbers only. If you are unsure between 1 and 2 on pattern work, look at the calendar: if more than half your week is template-shaped output, score 2. No half points.
Does a high score mean I will lose my job?
No. The score records exposure, not a pink slip. It tells you how much of your week sits in task shapes that can be handed over, so you know which move to make next.
Can I retake the test?
Yes, and you should. Date each score and revisit in 90 days. The number tracks your week as it changes, not a fixed trait.
What if my employer has already mandated AI tools?
That changes the timeline, not the score. Two people with the same 4 can face different adoption speeds. If your employer has set productivity targets on an AI tool, make your 30-day move sooner.
Is this test based on BLS data?
No. The test is a self-assessment of your week. For occupation-level outlook and task content, check the US Bureau of Labor Statistics Occupational Outlook Handbook and O*NET directly.
The number on the page, not the headline in your feed
The score measures exposure, not a pink slip. Nothing about it is a prediction, and the only thing that moves it is an action taken this month, not another headline.
Pick the move that matches your band and put it on the calendar. Low 0-1: log tasks for two weeks. Moderate 2-3: add a review or strategy step to your most pattern-heavy task. High 4-5: shift one task to oversight or client-facing work within 30 days. Very high 6: test one task from an adjacent role this month.
Back to the coordinator at $62,000 and 55% drafting: that 4 of 6 was never a sentence handed down, only a description of where the week currently sits, and the week is the part that can be rewritten. A low score is a reason to keep logging tasks, not to relax. Spend the two minutes once, then spend the month on the move.
Score your week in two minutes
Open your calendar, write down three numbers, and put your band-appropriate move in next week's schedule. No account needed.