Researchers Maxim Massenkoff and Peter McCrory from Anthropic published a detailed economic study titled “Labor Market Impacts of AI: A New Measure and Early Evidence” on the Anthropic Research blog. In plain terms, it tries to answer the question everyone is quietly asking: is AI actually taking jobs right now? The short answer is: not yet in a dramatic way, but some early warning signs are starting to show up.
Here’s why you should take 10 minutes to read this one.
What the Research is Actually About
Anthropic’s team created something called “observed exposure”, a new way to measure which jobs are genuinely at risk from AI . Instead of just asking “could AI theoretically do this job?”, they layered in real-world data from Claude usage to see what tasks people are actually automating right now .
The gap between what AI could do and what it is actually doing turns out to be massive. For example, AI could theoretically handle 94% of tasks in Computer and Math jobs, but in reality, Claude is only covering about 33% of those tasks today . That gap is the most interesting part of this whole study.
The Jobs at the Top of the Risk List
The research found the three most exposed occupations are :
- Computer Programmers (75% task coverage)
- Customer Service Representatives (high API automation observed)
- Data Entry Keyers (67% coverage, primary tasks already being automated)
On the flip side, cooks, bartenders, motorcycle mechanics, lifeguards, and dishwashers have zero measured AI coverage . Physical, in-person work remains a safe harbour for now.
Who Are the Most Exposed Workers?
This part of the study is genuinely surprising. The workers most at risk of AI displacement are not who you might expect. They tend to be :
- More educated (graduate degree holders are nearly 4x more represented in high-exposure jobs)
- Higher paid (earning 47% more on average than low-exposure workers)
- More likely to be female and white or Asian
So AI disruption, if and when it comes, is going to hit white-collar, knowledge workers hardest. Not factory floor workers or hourly staff.
The Hiring Slowdown for Young People
Here’s the early warning sign worth paying attention to. While overall unemployment rates for high-exposure workers have not spiked, the study found that workers aged 22 to 25 are getting hired into AI-exposed roles at a noticeably slower rate . Job entry rates into exposed occupations dropped roughly half a percentage point, a 14% fall compared to 2022 levels .
That means the door is not slamming shut. It is just quietly getting harder to open for people who are just entering the workforce. If you are a recent grad in tech, finance, or data roles, this research is practically written for you.
Why You Should Read the Full Article
- It gives you a framework to think about AI risk, not just fear it
- The data is built on real Claude usage, not just theoretical models
- It is one of the few studies that is honest about its own limitations and promises to update findings over time
- It will help you have smarter conversations about AI and jobs, whether you are a student, a manager, or someone planning your next career move
You can read the full Anthropic research paper here: Labor Market Impacts of AI: A New Measure and Early Evidence
The bottom line is that AI has not caused a jobs crisis yet, but this research gives us the tools to spot one before it becomes undeniable. That is exactly the kind of forward-looking, grounded thinking worth bookmarking.
Just Being Resourceful! 
Leave a comment