A kid in Ohio turned down a computer science scholarship last fall. Chose a union electrician apprenticeship instead. His parents thought he was throwing his future away. His future scores a 1 out of 10 on AI exposure. Theirs scored higher than they'd like to admit.
That kid isn't an outlier anymore. He's ahead of the curve.
42% of Gen Z is now actively pursuing trades. Plumbing. HVAC. Electrical. Welding. The jobs everyone told them to avoid. The jobs that, when you run the actual numbers across 500+ occupations, cluster at the lowest end of AI risk on the entire spectrum.
This is not a feel-good story about working with your hands. It's a cold, data-driven reckoning with who built the wrong mental model of the future.
The Assumption That Got Expensive
For two decades, the advice was simple. Get a degree. Get a white-collar job. Stay away from trades. That's the path that leads to stability, prestige, upward mobility.
Here's what that advice missed: the skills AI finds easiest to automate are structured, pattern-based, and language-driven. Which describes most of what happens inside an office.
The skills AI finds hardest to automate involve physical manipulation in unpredictable environments, spatial reasoning in real-world conditions, and human trust in high-stakes moments. Which describes most of what happens on a job site.
The education paradox
Bachelor's degree holders average 6.7 AI exposure. Workers without degrees average 4.1. More education, in most cases, means more exposure. The credential protected you from the last recession. Not this one.
Jobs paying $100K+ average 6.7 on AI exposure. Jobs paying under $35K average 3.4. The salary premium that came with white-collar work is now partially a risk premium too. You're being paid more, and you're being exposed more.
That's not the trade you thought you were making.
What Blue-Collar Jobs Actually Score
When Andrej Karpathy published his 342-occupation analysis on March 15, 2026, the numbers confirmed what anyone willing to look had already suspected. The lowest-scoring jobs on AI exposure are almost entirely physical and relational.
-
Electricians: 1/10. Every job site is different. Troubleshooting in real environments requires physical presence, dexterity, and judgment that no current model can replicate remotely.
-
Plumbers and HVAC technicians: 0-2/10. The combination of spatial problem-solving, physical manipulation, and variable conditions creates a nearly unassailable moat.
-
Nurses: 2/10. Physical care, patient relationships, and real-time clinical judgment in unpredictable human situations. The empathy layer alone defeats current automation.
-
Physical therapists: 3/10. Touch, observation, adaptation to individual patient response. High-trust, high-stakes. Not going anywhere.
The pattern is consistent. Physical presence. Variable environments. Human trust. These three factors lower AI exposure scores reliably across every category we analyzed.
The trades aren't AI-proof because robots can't exist in the physical world. They're AI-proof because the physical world keeps changing. Every crawlspace, every junction box, every patient is different.
The Comparison That Changes Everything
Look at healthcare side by side. Not as a sector. As individual roles.
Surgeons score 3. Radiologists score 7. Same hospital. Same prestige. Very different futures.
A radiologist reads images and produces structured reports. That's exactly what AI does well. Pattern recognition on standardized inputs. It's already happening. 81% of physicians now use AI daily, up from 38% in 2023. For radiology, that number is higher and accelerating.
A surgeon operates in real-time, in a living body, with hands and judgment and situational awareness. The physical manipulation required is far beyond current robotic precision for most procedures. The liability and trust dynamics alone extend the runway dramatically.
Same sector, opposite trajectories
Surgeons: 3/10 exposure. Radiologists: 7/10 exposure. The difference is not education or salary. It's whether the core work happens in physical reality or in structured data.
Now apply the same lens to business. VP of Sales scores 6. The SDRs reporting to that VP score 8. The person at the top is safer than the people doing the structured, repeatable tasks below them. AI is climbing the ladder from the bottom. Not the top.
That's the second-order effect most people don't see coming. Your title doesn't protect you. What you actually do inside that title does.
The Timeline Most People Underestimate
Not all exposure is equal. The score tells you what's coming. The timeline tells you when to worry.
-
Score 9-10: disruption now. Medical transcriptionists are already seeing -8% job outlook. This isn't projection. It's current data.
-
Score 7-8: restructuring in 2-3 years. Software developers sit here. But note: their job outlook is +25%. High exposure doesn't mean elimination. It means transformation. Some people won't survive the transformation.
-
Score 5-6: pressure in 5+ years. Enough runway to adapt if you start now. Not enough to ignore it.
-
Score 0-2: durable for the foreseeable future. The trades. Physical care. Environmental services. These aren't legacy jobs. They're infrastructure.
Only 3% of jobs score 9-10. The real story is the 42% that score 7-8. That's 59.9 million jobs and $3.7 trillion in wages that face meaningful restructuring in the next 2-3 years. Not elimination. Restructuring. The distinction matters enormously for what you should do about it.
Where does your job actually land?
500+ occupations scored 0-10. Free. Takes 60 seconds.
Why Gen Z Read This Correctly
Gen Z didn't stumble into the trades. They ran the numbers, consciously or not, and made a bet. The bet is looking smarter by the quarter.
Consider the full picture of what they're choosing. An electrician apprenticeship: 4-5 years, minimal debt, $0 tuition in most union programs, a license with legal protection, and a score of 1 on AI exposure. Day one of licensure, they're working on something no algorithm touches remotely.
Compare that to the person who spent $120,000 on a communications degree to enter a marketing coordinator role that scores 7 on AI exposure and is already being restructured by automation tools. The math isn't close.
Gen Z's strategic pivot
42% of Gen Z pursuing trades with plumbing and HVAC scoring 0-2 on AI exposure. They looked at the data, even if they didn't see the data. The instinct was right.
There's also something the pure AI-exposure score doesn't fully capture: the AI skills premium. People who work in high-exposure fields and develop AI proficiency earn a 56% salary premium. That's real. But it requires surviving the transition. In trades, you don't need to survive a transition. You're already on the right side.
The savviest move? A tradesperson who adds AI literacy on top of physical expertise. Scheduling optimization. Automated quoting. Predictive maintenance modeling. Low exposure core. AI-augmented periphery. That's the combination that's hard to compete with.
What to Actually Do With This
Whether you're in the trades, eyeing them, or deep in a white-collar role that scores higher than you'd like, the same framework applies. Know your score. Understand your timeline. Act before the window closes.
Find your actual exposure score, not just your job title's score. A software developer title covers people who architect systems (lower exposure) and people who write boilerplate code (higher exposure). The tasks matter more than the title. Look at what you do hour by hour.
Identify the physical or relational layer in your work. Every role has some. Client relationships. Physical presence. Judgment calls that require being there. Lean into those. They're your natural hedge.
If you're at 7+, develop AI fluency in the next 12 months. Not to fight the automation. To work with it. The 56% salary premium for AI skills goes to the people who become the interface between human judgment and AI output. That role is durable. Resisting the tools is not.
If you're evaluating a career change, run the numbers on trades seriously. Electricians. Plumbers. HVAC. Welders. The earnings gap with white-collar work has narrowed substantially. The debt burden comparison is stark. The AI exposure comparison is decisive.
The full occupation breakdown covers 500+ jobs with exposure scores, timeline estimates, and specific restructuring signals. This article covers the pattern. The report covers your specific role.
The global average AI exposure is 5.3. That means most workers are sitting in the middle of a risk curve that is actively moving. Blue-collar workers at 0-2 are not in the middle of anything. They're on the far safe end, and the distance between them and the average is growing.
Bottom line: The jobs we told a generation to avoid are the jobs best positioned to survive what's coming. The prestige hierarchy we built around credentials and office work is colliding with an AI transition that rewards physical reality over structured information. That Ohio kid with the electrician apprenticeship isn't behind. He just read a map the rest of us were handed upside down.
Find out where you stand
500+ occupations scored 0-10 on AI displacement risk. Free.