Is My Job at Risk From AI? A Simple Self Check Built on Anthropic's Task Model

Anthropic's Economics team has published a model that gives you a much better way to answer this question than reading another list of endangered professions. The model, laid out in Economic Scenarios for Transformative AI (Korinek et al., 2026), does not sort jobs into safe and unsafe. It breaks every job into tasks and asks what happens to each one.

That is the right unit, and it means you can run the analysis on yourself. Today. In about twenty minutes.

I am Riten, founder of Fueler, a platform where people get hired through their work rather than their resume. I have run this exercise with a lot of people now, and it does something that no prediction article can. It replaces a vague dread with a specific list.

Here is how to do it.

The Framework Anthropic Uses, in Plain Language

The model treats every job as a bundle of tasks. It uses the US Department of Labor's O*NET taxonomy, which lists the tasks that make up each occupation. Then each task falls into one of four groups.

Untouched. AI cannot do it. The report's example is blunt: AI cannot bathe a patient.

Augmented. AI helps a human do it better or faster. For their nurse, that includes drafting discharge instructions, monitoring patients remotely and planning the shift's care schedule.

Automated. AI takes it over entirely. For the nurse, that might be charting a patient's vitals or ordering supplies for the ward.

New. A task that exists only because AI does. Checking how well an AI triaged a patient, or reviewing an AI proposed care plan.

The nurse's job title never changes. What changes is the proportions. And the report notes that the nurse ends up able to oversee and accomplish more, spending more of her time talking with patients and helping them understand their diagnoses.

That is the actual shape of the change for most people. Not deletion. Redistribution.

Five Steps to Check If Your Own Job Is at Risk

Get a blank sheet of paper. This works better on paper than in your head, because your head will skip the uncomfortable ones.

• Step one: write every task you actually do. Not your job description, which is usually fiction. What you really spend hours on in a normal week. Most people find between eight and fifteen tasks. Be specific. "Write reports" is too vague. "Pull last month's numbers, build the deck, present it to the regional team" is three separate tasks.
• Step two: mark each task with one of the four labels. Untouched, augmented, automated, or new. Be honest on the automated ones. The instinct is to protect your own work, and that instinct will make the exercise useless. If a tool can already do it at 80% quality, mark it.
• Step three: count the hours, not the tasks. Five untouched tasks that take two hours a week are worth less protection than one automated task that takes twenty. Weight your list by time. This is where most people's picture changes sharply.
• Step four: look at the new column. If it is empty, that is your most important finding. The report notes that historically new technology creates new tasks for workers, and in Anthropic's extreme scenario the danger is precisely that AI creates essentially no new knowledge tasks. If nothing new has entered your bundle in two years, you are drifting rather than adapting.
• Step five: ask which of these anyone can see. For every task still clearly yours, ask whether a person hiring you could verify it. If the answer is no across the board, the risk is not that your job disappears. It is that you cannot prove your value when you need to.

Why It Matters

Running this on yourself is better than reading forecasts, because the forecasts are about averages and you are not an average.

But the forecast context is still useful, and it is less frightening than most headlines suggest.

In Anthropic's modest scenario, AI has roughly the impact the internet had. In the substantial scenario, AI is capable of half of all knowledge work by 2030, yet most knowledge work is still done without AI because adoption trails capability. In both of those, job reallocation and unemployment stay within ranges history has already seen. Only in the extreme scenario does unemployment spike beyond typical recessionary levels.

Anthropic surveyed more than 10,000 Americans in August, and the typical respondent's answers implied something close to the substantial scenario, with unemployment around 5% by 2030. Roughly 10% held views matching the extreme case.

The wage picture is where the real pressure sits. Average wages rise in all three scenarios, but the gains concentrate outside knowledge work. Knowledge worker wages are essentially flat in the substantial scenario and fall more than 10% by 2030 in the extreme one.

So for most people, the honest risk is not unemployment. It is being unable to justify your rate because nobody can see which tasks are still yours.

The report is also open about its limits. It excludes policy responses, business cycles, financial market disruptions and hyper capable robots, and it does not follow individual workers. Reviewers disagreed with the authors in both directions.

What This Means for Your Portfolio and How You Get Hired

Step five above is the one I care about most, and it is the one people skip.

You can have a beautiful list of tasks that are genuinely still yours and still lose the job to someone worse, because the person hiring cannot see any of it. Ability that nobody can verify is worth almost nothing in a hiring process.

This is the entire reason I built Fueler.

Turn your untouched and new tasks into documented projects. Take the tasks that survived your audit and build a page around each one showing a real example. What the problem was, how you approached it, what you decided. That is the substance of a proof of work portfolio built step by step.

Write your augmented tasks honestly, including the AI part. If AI now does the first draft and you do the direction and the edit, say exactly that. On Fueler you can add your AI stack to a project and lay the workflow out clearly. Adoption is lagging capability across the whole economy, so proving you have a real workflow puts you ahead, not behind.

Fill the empty new column deliberately. If nothing new has entered your bundle, go create something. Build a review process, design a workflow, take on the checking work nobody has claimed yet. Then document it. New work is the most valuable thing in a portfolio precisely because so few people have it written down.

If you are just starting, run the audit on the job you want. You do not need a job to do this exercise. Look at the role you are aiming for, break it into tasks, and build proof against the ones that survive. The student guide to building a first portfolio shows how to do it from zero.

Use a structure so you finish. Most people quit at the blank page. The 6 step portfolio formula gets you from nothing to published without stalling.

Final Thoughts

The question "is my job at risk from AI" cannot be answered by anyone but you, because nobody else knows what is actually inside your week.

What Anthropic has given us is not an answer. It is a better method. Stop thinking about your job as a single thing that either survives or does not. Think about it as eight to fifteen separate pieces of work, each with its own future.

Do that and something useful happens. The fear becomes a list. Lists can be worked on. Some items you protect, some you hand over gladly, some you go and create.

Then do the part almost everyone forgets, which is making the surviving work visible to somebody who might hire you. That is the difference between having value and being able to prove it, and in a market under pricing pressure, only one of those pays.

Run the audit this week. Then go build a career portfolio that actually gets jobs around what is left.  

Frequently Asked Questions

Is my job at risk from AI?

Break your job into its individual tasks and label each one as untouched by AI, augmented by AI, automated by AI, or newly created because of AI. Anthropic's model treats risk as a task level question rather than a job level one, so a task by task audit gives a far more accurate answer than any list of endangered professions.

How do I know which of my tasks AI can do?

Test them. Take each task on your list and try to complete it with an AI tool honestly, then judge the output at the standard your work is actually held to. The instinct to protect your own work will distort this, so weigh the result by how much time the task takes rather than how important it feels.

Will AI take my job completely?

In Anthropic's modest and substantial scenarios, job reallocation and unemployment stay within ranges history has already seen, and most knowledge work is still performed without AI even when AI is capable of it. Full job replacement is concentrated in the extreme scenario, which would likely require recursively self improving AI adopted very quickly.

What should I do if most of my tasks can be automated?

Focus on the tasks that survived, deliberately take on new work such as reviewing or directing AI output, and build visible evidence of both. The report notes that switching occupations is slow and difficult, so building proof while you are still employed is far easier than doing it under pressure.

How do I prove my value to employers in the AI era?

Document your work at the task level, explain the decisions behind it, show how you use AI within your process, and publish it where a hiring manager can verify it. Evidence of judgment is what remains valuable when the production of output becomes cheap.

Why 100,000+ professionals use Fueler

Fueler helps professionals showcase proof of work through projects, assignments, case studies, and achievements.

• Thousands of professionals use Fueler to create their digital portfolio
• Thousands of projects are published on Fueler. Check here
• Startups and Companies hire through proof of work on Fueler
• Used by freelancers, creators, marketers, video editors, writers, designers, and product managers

Our mission is to help the next 100 million professionals build a verified professional identity through proof of work