Anthropic Just Mapped Which Jobs AI Is Actually Replacing. Here's What It Means for Your Search.

The number: 75%

Anthropic released a new study last week measuring which jobs AI is actually displacing. Not in theory. In observed, real-world usage.

Their metric, "observed exposure," combines three things: whether a task is theoretically possible with an LLM, whether people are actually using AI to do that task, and whether that usage is automated (AI doing the work) versus augmentative (AI helping a human do the work). Automated use counts more.

The most exposed occupation: computer programmers, at 75% task coverage. That means three out of four tasks in that role are already being done by AI in real professional settings.

The top 10

The study ranks roughly 800 US occupations. Here are the most exposed:

• 🔴 Computer Programmers: 75%. Three-quarters of their tasks show up in automated AI usage. This isn't projection. It's what's already happening.
• 🔴 Customer Service Representatives. Main tasks increasingly handled through API-based automation. First-party, not human-assisted.
• 🔴 Data Entry Keyers: 67%. The primary task of reading source documents and entering data sees significant automation.
• 🔴 Financial Analysts, Technical Writers, Market Research Analysts. All in the top 10. Knowledge work roles that depend heavily on information processing, synthesis, and drafting.

At the other end: 30% of all US workers have zero observed exposure. Their tasks appeared too infrequently in the data to register. Cooks. Motorcycle mechanics. Lifeguards. Bartenders. Dressing room attendants.

The gap between "what AI could do" and "what AI is doing" is still enormous. Computer & Math occupations are 94% theoretically exposed but only 33% covered in actual usage. That gap is your window. But it's closing.

The demographic surprise

This is the finding that should change how you think about AI risk.

The workers most exposed to AI are not the ones you'd expect if you've been reading headlines about automation replacing factory workers. They're:

• 16 percentage points more likely to be female than the least exposed group
• 11 percentage points more likely to be white. Almost twice as likely to be Asian
• Earning 47% more than the unexposed group on average
• Far more educated. 17.4% of the most exposed group holds a graduate degree, compared to 4.5% of the unexposed group

This is not a blue-collar disruption story. The workers with the highest AI exposure are educated, experienced, and well-paid. That's the profile of almost everyone reading this. And it's happening in a labor market that lost 92,000 jobs in February with no leading sector pulling things forward.

The young worker signal

There's one finding in this study that hints at where things are heading.

Unemployment hasn't increased for workers in the most exposed occupations. Not yet. The researchers looked hard and found no statistically significant change since late 2022.

But hiring has slowed for young workers. Workers aged 22-25 are 14% less likely to start a new job in a high-exposure occupation compared to 2022. That rate has been flat for less-exposed occupations.

Why this matters to you: when companies stop hiring juniors, the work doesn't disappear. Seniors absorb it. Or AI handles it. Either way, the expectations of your role shift without the title or comp changing to match. And if the pipeline of junior talent dries up, the teams you manage get thinner.

What this means for your search right now

• Know which of your tasks are exposed. Not your job title. Your tasks. If 60% of what you do is information synthesis and drafting, you need to know that. If the other 40% is judgment, relationship management, and cross-functional leadership, that's your leverage. Lead with it.

• "How does your team use AI?" is now a critical interview question. Not as a culture test. As a role-definition question. If the company has already automated the tasks you'd be doing, you need to know before you accept. If they haven't, you need to know whether they expect you to.

• Watch for the junior-hiring signal at companies you're considering. If a company has stopped backfilling junior roles on a team, that tells you something about how they view the function. Ask: "What does your headcount plan look like for this team over the next 12 months?" and "Are you hiring at other levels for this function?"

• The gap between theory and reality is your strategic advantage right now. 94% of Computer & Math tasks are theoretically automatable. Only 33% are being automated today. The people who learn to work alongside AI in the remaining 67% will be more valuable, not less. But only if they're intentional about it.

• Augmentation counts. The study distinguishes between automation (AI replacing the task) and augmentation (AI assisting). Right now, most usage is augmentative. The workers who thrive will be the ones who can clearly articulate: "Here is exactly how I use AI, here is what it produces, and here is the judgment I add on top." That's a job skill now. And in a market where candidates are paying thousands just to get noticed, it's a better investment than mass-applying.

How Kinship factors this in

The roles you're looking at aren't the same roles they were two years ago. A "Senior Financial Analyst" posting in 2026 is a fundamentally different job than it was in 2023, even if the title, the pay band, and the job description haven't changed.

Every week, we scan companies in the Kinship network for AI-driven restructuring, layoffs, hiring freezes, and shifts in team composition. When a company announces it's cutting roles because of "AI-driven efficiency" - like Block (40% headcount), HP (4,000-6,000 roles), and Wise (29% of workforce) did in the last two weeks - that's a signal we factor into your job scores.

But it's not just about avoiding risk. Companies investing in AI are also creating new kinds of roles. The ones with fresh capital and growing teams need people who can work at the intersection of domain expertise and AI tooling. Kinship surfaces those opportunities too.

The research is clear: the most exposed workers are educated, experienced, and well-paid. If that's you, you don't need to panic. You need better information. That's what we're building.