The Number That Breaks the Model
A senior engineer at a mid-sized SaaS company told me something last month that stuck. He'd just read that software developers score 8 out of 10 on AI exposure. His first thought: "I'm cooked." His second thought, when he checked the BLS outlook: "Wait, +25% growth through 2032?"
Both facts are true. That's the problem.
Most conversations about AI tech jobs collapse into one of two camps: "AI will take everything" or "AI just creates more jobs." Neither camp is wrong. Both are incomplete. The data tells a more precise, more uncomfortable story.
A 10-point exposure score doesn't mean your job disappears. It means your job transforms. And transformation has winners and losers inside the same title.
What You're Getting Wrong About High Scores
Andrej Karpathy published a 342-occupation analysis on March 15, 2026. The headline numbers were jarring. But the nuance buried inside them is what most people missed.
A score of 9-10 means disruption is happening now. A score of 7-8, where most software developer AI growth conversation lives, means restructuring over 2-3 years. Not elimination. Restructuring.
The Score Breakdown
Only 3% of occupations score 9-10. That's near-full automation, happening right now. The bulk, 7-8, means your job is being restructured, not erased. The timeline is 2-3 years, not 2-3 months.
The distinction matters. A medical transcriptionist scores 10. Job outlook: -8%. That's the danger zone. A software developer scores 8-9. Job outlook: +25%. Same high score. Opposite trajectories.
The variable that separates them isn't exposure. It's what happens to demand when AI touches the work.
Transcription is a cost center. Cheaper AI means less need for humans. Software development is a capacity constraint. Better tools unlock more projects, more products, more engineers hired to ship them. High score. Divergent demand. The score alone tells you almost nothing.
The Salary Trap Nobody Is Talking About
Here's the one that makes people uncomfortable.
You think your degree protects you. You think your salary signals safety. The data says the opposite. Jobs paying $100K+ average an AI exposure score of 6.7. Jobs under $35K average 3.4. The higher your income, the more of your work is already automatable.
The Education Paradox
Bachelor's degree holders average a 6.7 exposure score. People without a degree average 4.1. Four years of education and premium salary, and you're actually more exposed. Not less.
The plumber isn't worried. The HVAC technician isn't worried. They score 0-2. They work in physical space, with their hands, in non-repeatable conditions. AI can't unblock a drain or replace a condenser coil. 42% of Gen Z is already routing toward trades, and this data tells you why.
Meanwhile, the software developer AI growth story hides a second-order effect that's more immediate. Look one level below the title.
A VP of Sales scores 6. The SDRs under them score 8. The tasks AI takes first are the ones at the bottom of the org chart: the outreach, the qualification, the pipeline building. The manager's job survives. The entry-level path into that manager job gets restructured before you ever get there.
The junior role is often where AI lands first. The career ladder doesn't disappear. The bottom rungs do. That's a different problem entirely.
Same Industry, Different Futures
Healthcare is the clearest proof that the title is almost irrelevant. The task mix is everything.
Nurses score 2. Surgeons score 3. They work with patients, bodies, real-time complexity that resists automation. Their jobs are not going anywhere. But radiologists score 7. The core of that work, reading images and pattern-matching against known conditions, is exactly what AI does well.
81% of physicians now use AI daily. Up from 38% in 2023. That adoption rate didn't replace physicians. It restructured what they spend their time on. But for radiologists, that restructuring cuts closer to the core competency than for almost anyone else in medicine.
Healthcare Score Spread
Nurse: 2/10. Surgeon: 3/10. Radiologist: 7/10. Same hospital. Same building. Three completely different futures. Your industry doesn't protect you. Your task mix does.
The same logic applies in tech. A software developer AI growth narrative gets treated as monolithic. But a developer who writes boilerplate CRUD endpoints at a legacy bank scores very differently in practical terms than a developer who architects distributed systems and navigates ambiguous product requirements.
The score is an average across task types. Your actual exposure depends on which tasks consume your day.
What You Can Actually Do About It
The global average AI exposure score is 5.3 out of 10. That's not a crisis number. It's a pressure number. Enough to reshape careers. Not enough to eliminate most of them. The window is still open. But it's narrowing at different speeds for different roles.
The tech career AI question isn't "will I be replaced?" It's "what does my role look like in 3 years, and am I building toward that version or the version that gets restructured away?"
Here's what the data points toward, practically:
-
AI skills command a 56% salary premium right now. The leverage is available to anyone willing to develop it. That premium exists because supply of AI-fluent workers is still behind demand.
-
Score 7-8 means 2-3 years of runway. Not urgency measured in weeks. Urgency measured in the career decisions you make this year and next. That's still actionable time.
-
Score 9-10 is a different conversation entirely. Medical transcription, basic data entry, templated writing. If your core value is doing the thing AI now does faster and cheaper, the runway is shorter than 2 years.
-
Move toward judgment, not just execution. The tasks that survive automation are the ones that require context, ambiguity, accountability. The developer who can architect. The analyst who can present to the board. The clinician who makes the call.
Where do you stand?
500+ occupations scored 0-10. Free. Takes 60 seconds.
The Misread That's Costing People
The most common mistake: looking up the score for your job title, seeing a 7 or an 8, and assuming you know where you stand. You don't. Not yet.
42% of US jobs score 7 or higher. That's 59.9 million jobs and $3.7 trillion in wages in the restructuring zone. The score is a distribution, not a verdict. The median developer, the median analyst, the median financial advisor, they're all in that 7-8 band. That doesn't mean they all face the same future.
Your score is a starting point, not a sentence. What you do in the next 18 months is the variable the score can't capture.
The survival playbook isn't complicated in concept. It's just uncomfortable in execution. Move toward the tasks inside your role that require human judgment. Get fluent in AI tools as a force multiplier, not a threat. Understand the second-order effects in your org chart, not just the first-order exposure of your title.
The full analysis, covering 12 specific actions by occupation category, lives inside the survival report. This article gives you the frame. The specifics go deeper.
Bottom Line
Software developers score 8-9 on AI exposure. They also face +25% job growth. Both facts are true. The paradox resolves when you understand that high exposure plus expanding demand equals restructuring, not elimination.
The score tells you how much AI touches the work. It doesn't tell you what happens to demand for the work when it does. Those are two completely different questions.
Medical transcriptionists score 10 and the outlook is -8%. Software developers score 8-9 and the outlook is +25%. Same high exposure. Opposite futures. The difference is whether AI is a cost-cutter or a capacity-expander for the work you do.
Know your score. Then ask the harder question: when AI handles more of the routine work, does the world need more or fewer people doing your job?
The score is the map. The demand question is the territory.
Find out where you stand
500+ occupations scored 0-10 on AI displacement risk. Free.