Blog

How to Use AI-Risk Ratings When Choosing a Career: Seven Suggestions

JJesse Orrico, founder, Tarva, LLC·November 9, 2026

Once people have seen the AI ratings in Career Universe, the next question is a practical one: so what do I do with them?

Here are seven suggestions. They're how I would use the map, whether you're a student, a parent helping one, or someone changing careers. They are ideas to consider, not instructions. The ratings underneath them are our estimates, and the people who know your situation best are you and the people around you.

Suggestion 1: fit first, risk second

A low rating doesn't make a job right for you. If you're afraid of heights, it doesn't matter that our estimate for wind turbine technician is low at 10 years. You'd be climbing 300 feet up a tower.

So start with what you're like. Every job card has an "Is it you?" section: the kind of person who tends to enjoy it, whether it's indoors or outdoors, how physical it is, the usual hours, and how often people work remotely. In the Filters panel, "Kind of person" lets you pick from Hands-on builder, Curious thinker, Creator, Helper, Leader and Organizer.

Find a handful of jobs that sound like you. Then look at the risk.

Suggestion 2: read the 10-year rating if you're in high school

Every career has two ratings, one for the next 5 years (through about 2031) and one for the next 10 (through about 2036).

If you're 15, the 10-year rating is closer to your real starting line. Depending on the path, you might finish training somewhere in that window.

The two pictures are different. In our estimates, 62% of careers are rated low risk for the next 5 years. At 10 years, it's 44%. Part of that shift is robots and self-driving machines, which our rubric expects to arrive slower than software.

If you're changing careers and could start in months, read both. The 5-year rating describes your first years in the job. The 10-year rating describes the years after that.

Suggestion 3: treat "medium" as "changing", not "doomed"

About 39% of the careers in the map are rated medium at 10 years. That's a lot of the map to cross off, and I don't think you should.

Medium means a real share of the tasks gets automated, with fewer or different openings, especially at entry level, but the role stays and changes. The one-sentence "how" tells you where the change lands. Two examples from the map:

• Software developer, medium and medium: "AI coding agents write and test much routine code, so fewer junior hires; designing systems and owning outcomes stay human."
• Accountant, medium and medium: "AI automates reconciliations and draft statements, trimming entry-level work; accountants who interpret, advise and sign off remain needed."

Notice the pattern. The squeeze is on the beginner's work. So the useful question for a medium job isn't "should I avoid it?" It's "how would I get past the part AI is taking?" Sometimes the answer is a different first job, a certification, or a path that starts you closer to the human part.

To read the sentence on any job, open it and tap "Why" under "Will it last?".

Suggestion 4: stack the filters

The quick filters work together. Turn on more than one and you see only jobs that match all of them. Here's how the map narrows when you stack a few:

A few other stacks worth trying:

• AI-safe for 10 yrs and Hidden gems: 1,900+ careers most people have never heard of.
• AI-safe for 10 yrs and Get paid to train: 1,500+.
• AI-safe for 10 yrs and Typically $100k+: 1,100+.

These counts come from our estimates, and the "AI-safe" label is our shorthand for "rated low risk". It isn't a promise.

One more thing. Stacking is great for finding ideas, but the last row shows the catch. Pile on enough filters and the menu gets small again. When it does, turn one off.

Suggestion 5: look for jobs where AI is the tool, not the worker

In our ratings, "low" means AI and robots look like tools in the job, and human demand holds steady or grows. That distinction, tool or worker, shows up in the research too.

The Stanford Digital Economy Lab's "Canaries in the Coal Mine?" paper found that the drop in employment for young workers was concentrated in jobs where AI automates the work rather than helps people do it. The Anthropic Economic Index found that about 57% of the use it saw looked like augmentation, a person working with the AI, and 43% like automation. (That one measures how people use a single company's AI assistant, not the job market, so treat it as a clue.)

When you read a "how" sentence, ask one question: is AI taking the task, or helping with it? Two examples where it's helping:

• Registered nurse: "AI charting and early-warning alerts cut paperwork; bedside care, judgment and patient advocacy stay human, and demand keeps growing."
• Dental hygienist: "AI flags gum disease and cavities on X-rays; scaling, cleaning and coaching patients chairside stay hands-on."

If you'd like to go deeper on the research, How Will AI Affect Jobs? walks through it in plain language.

Suggestion 6: start something this year

A career idea gets real when you try a piece of it. The map has three places to look.

• Start today (a quick filter, 1,100+ careers) marks jobs where someone could realistically begin an entry-level version within a few days, with no course, license or test first. Each one shows the youngest age that's usually allowed and one way to start.
• Could start within a month shows up on 400+ more job cards, where there's usually one short step first.
• Start now, in high school is on every job card, with classes that help and one to three things a teenager could try now: a club, a starter job, a certificate.

If you're an adult, skip the high-school section and spend time with Get paid to train and Working within 2 yrs instead.

Pick one thing and put it on the calendar this month. Small is fine.

Suggestion 7: check with a human

This is the one I'd keep if I could only keep one. The map gives you a list. People who do the work give you the real picture.

For each job on your shortlist, find one person who does it. A relative's friend, a local business, a professional association, a teacher who knows someone. Ask them:

• What did a beginner do here five years ago, and what do they do now?
• Is AI changing who you hire, or how many?
• What would you tell someone starting today?

Then check anything important with official sources: the BLS Occupational Outlook Handbook, the school's own pages, and the licensing board in your state. The government's own projections are a useful second opinion. The BLS Employment Projections for 2024–34 show healthcare and social assistance as the largest and fastest-growing sector.

What we wouldn't do

• Pick a career only because it's rated low. Fit and pay and training cost all matter at least as much.
• Cross off something you love because it's medium or high. Read the "how", then talk to people in the job. A high rating is a reason to look closer, not a verdict.
• Take on big training costs to chase a rating. The ratings are estimates. Some of them are probably wrong, and we update the data from time to time.
• Treat the counts as forecasts. They describe how we applied our rubric, not what the economy will do.

The ratings were written with AI research assistants against a strict schema, checked by validation scripts and spot-checked by hand, by a very small team. How we estimated AI risk explains the method. If one looks wrong to you, email privacy@tarvacore.ai with the job title, the ratings you'd give and one sentence on why.

Questions people ask

Should I avoid careers rated high risk?

We wouldn't choose or rule out a career on its rating alone. Fit comes first. Then read the 10-year rating and, more importantly, the one-sentence reason saying what changes and what stays human. A high rating is a reason to look closer and talk to people in the job, not a verdict.

Which rating should a 15-year-old look at?

Both, but give the 10-year rating more weight. A 15-year-old is likely to be starting work somewhere between about 2031 and 2036, which is the window the 10-year rating covers.

What does "Get paid to train" include?

Jobs where the usual path pays you while you learn: registered apprenticeships, military training, and paid on-the-job training such as police and fire academies or utility lineworker programs. Pay during training is often modest. Each job explains the typical route.

What does "Start today" mean?

It marks jobs where someone could realistically begin an entry-level version within a few days, with no course, license, certificate or test first. That means being hired and trained on the job, or starting on your own with under about $500 of gear. Each one shows the youngest age at which that's usually allowed and a first step. Rules vary by state, so check locally.

How do I save a shortlist?

Open any job and tap Add to shortlist. As a guest, your stars and filter settings stay in your browser on that device. If you want the same shortlist on another device, you can create a free account with an email code and choose which stars to sync.

Your next step: open the map and tap AI-safe for 10 yrs and No 4-year degree (or any filter that fits you). Star five jobs. Then turn both filters off and star one more you simply like, whatever its rating.

Sources: Brynjolfsson, Chandar and Chen, "Canaries in the Coal Mine? Six Facts about the Recent Employment Effects of Artificial Intelligence", Stanford Digital Economy Lab, 2025, revised August 2026. Anthropic, The Anthropic Economic Index, 2025. US Bureau of Labor Statistics, Employment Projections 2024–34, August 28, 2025, and the Occupational Outlook Handbook. Counts and shares are from our own estimates in the map. Parts of the research behind Career Universe were written with AI research assistants and checked by hand.