How AI Is Impacting the Job Market (So Far)
New research from Anthropic tracks what AI can theoretically do versus what it's actually doing at work. There's a massive gap between the two. But the data on which workers are most exposed, particularly women, challenges the usual assumptions.
Plenty of people have opinions about AI taking jobs. Most of those opinions are based on what AI could do, not what it's doing right now. Anthropic's latest research tackles that head on.
Measuring What's Actually Happening
The research team built a new metric called "observed exposure." It combines theoretical AI capability with real world usage data. Basically, it measures the difference between what's possible on paper and what's happening in practice.
At the moment, that gap is large. AI could theoretically handle 94% of tasks in Computer and Math occupations. The actual number today? 33%.
The Jobs Seeing the Most AI Activity
Three occupations top the list:
- Computer programmers at 74.5%
- Customer service representatives at 70.1%
- Data entry keyers at 67.1%
Women Are More Exposed Than Men
The demographic breakdown challenges the usual narrative about technology displacing lower wage workers. Workers in the most exposed roles are 16% more likely to be female compared to those in unexposed jobs. They're also older, more educated, and higher paid.
This matters because policy conversations about AI displacement tend to focus on blue collar and lower income workers. The data suggests the impact could hit a different population entirely.
So Is AI Causing Unemployment?
According to Anthropic, not yet. The data shows no real increase in unemployment for exposed occupations since late 2022.
But something is happening with younger workers. Hiring rates for people ages 22 to 25 appear to be slowing in these exposed fields. The researchers are careful to note this isn't definitive. Those workers might be staying put, switching careers, or going back to school. But it's a trend worth watching.
Why You Should Care
The paper says that labor market predictions are often wrong. A decade ago, experts said 25% of US jobs were vulnerable to offshoring. Most of those jobs ended up growing just fine.
This research takes a different approach. By setting up measurement frameworks now, before any real disruption, it becomes easier to spot actual changes when they happen.
The full report goes deeper into the methodology and breaks down the demographics. If you want to know what the numbers actually say instead of what the headlines claim, it's worth your time.