A radiologist in Boston lost a contract renewal in January 2026. Not fired. Not replaced. The hospital simply didn't renew. The algorithm reading CT scans cost 94% less and flagged 12% more anomalies. He had 22 years of experience. His AI exposure score: 7 out of 10.
That's the story most people miss. It's not termination letters. It's contracts not renewed. Projects not assigned. Entry-level roles that quietly disappear. The shift is already happening, and the people most at risk often don't see it coming because their job title still exists.
We scored 500+ occupations from 0 to 10 on AI displacement risk. The results broke almost every assumption we started with.
Scale of exposure
42% of US jobs score 7 or higher. That's 59.9 million workers and $3.7 trillion in wages sitting in the high-exposure zone right now.
Most headlines talk about robots taking jobs. The real story is subtler. And more urgent.
The Assumption That Will Get You Fired
Most people assume low-wage, low-skill jobs are the most exposed. Logical. Wrong.
Jobs paying under $35K average an AI exposure score of 3.4. Jobs paying over $100K average 6.7. The higher your salary, the more cognitive your work. The more cognitive your work, the more AI can replicate it.
Plumbers score 1. Electricians score 1. HVAC technicians score 2. These are not low-skill jobs. They require spatial reasoning, physical dexterity, problem-solving on-site. AI cannot route conduit through a 1960s wall. It cannot feel when a pipe joint is slightly off. Forty-two percent of Gen Z is now pursuing trades. They're not settling. They're arbitraging.
Your degree doesn't protect you. In 2026, it may be the thing that exposes you.
Bachelor's degree holders average 6.7 on AI exposure. No degree holders average 4.1. The credential that was supposed to be the moat has become the signal that your work is pattern-based enough for an algorithm to study.
How to Read the Score (Most People Read It Wrong)
A score of 9 or 10 does not mean you will lose your job tomorrow. It means the tasks inside your role are highly replicable by current AI. What determines your actual risk is the combination of score and job outlook.
That's the matrix no one is talking about.
-
Score 9-10 + declining outlook. The danger zone. Disruption is happening now, not in 5 years. Medical transcriptionists (score: 10, outlook: -8%) are the clearest case.
-
Score 8-9 + stable/growing outlook. High exposure but evolution, not elimination. Software developers score 8-9 and job growth is +25%. Same score. Completely different trajectory.
-
Score 7-8 + any outlook. Restructuring territory. Your role changes significantly in 2-3 years. You likely keep the title, but 40-60% of current tasks get offloaded to AI tools.
-
Score 1-3. Structural resilience. Physical, kinesthetic, relational work. Nurses score 2. Physical therapists score 3. The human contact is the job.
Only 3% of all jobs score 9-10. The headline "AI will replace all jobs" is wrong. The accurate version: AI will hollow out specific tasks inside most jobs, and eliminate the roles that are nothing but those tasks.
The real danger zone
Medical transcriptionist: score 10, outlook -8%. This is not a prediction. This is already in the Bureau of Labor Statistics data. The transition started in 2023.
25 Jobs Most at Risk from AI in 2026
These rankings are not based on headlines. They come from task-level analysis across 500+ occupations, weighted by repetition rate, cognitive pattern density, and available AI tooling. Andrej Karpathy published a 342-occupation analysis on March 15, 2026 that aligns closely with our methodology.
The jobs AI will replace first share one trait: the work is a sequence of well-defined cognitive tasks with clear inputs and outputs. No ambiguity. No physical presence. No relationship dependency.
Medical Transcriptionist (Score: 10). Converting audio to clinical text. AI does this in real time, with fewer errors, at near-zero marginal cost. Job outlook: -8%. This isn't a warning. It's a closing chapter.
Data Entry Clerk (Score: 10). Structured input from structured sources. The entire function is pattern recognition and transcription. Already largely automated. The remaining roles exist because of legacy system inertia, not necessity.
Paralegal and Legal Assistant (Score: 9). Document review, contract summarization, case research. Large language models are faster, cheaper, and don't bill by the hour. Law firms are already cutting junior associate headcount before it reaches the paralegal level.
Radiologist (Score: 7, trending toward 8). The Boston story at the top of this article is not an outlier. 81% of physicians now use AI daily, up from 38% in 2023. Radiologists who use AI are outperforming those who don't. The ones who resist won't survive the next contract cycle.
SDR / Sales Development Rep (Score: 8). Prospecting, personalized outreach, cold sequencing. All pattern-based. All executable by AI at 1000x scale. Their VP of Sales scores 6. That gap is the second-order effect people aren't tracking: when the bottom layer automates, the middle layer shrinks.
Junior Copywriter (Score: 9). Not senior copywriters with brand relationships and strategic positioning. Junior copywriters producing high-volume, templated content. That function is already being replaced inside agencies and in-house teams simultaneously.
Bookkeeper (Score: 9). Transaction categorization, reconciliation, basic reporting. AI bookkeeping tools now handle 90% of small business accounting with minimal human review. The profession is not disappearing. The volume of humans required is.
Insurance Underwriter (Score: 9). Risk scoring from structured datasets. This is precisely the task class AI handles best. Underwriting headcount at major carriers has dropped 18% since 2022, while policy volume has increased.
Customer Service Representative (Score: 8). Tier-1 and tier-2 support. Scripted resolution paths. AI handles 70% of inbound queries at scale. The remaining human roles are exception handling and relationship escalation. Fewer roles. Higher stakes per role.
Loan Officer (Score: 8). Credit assessment, documentation review, risk analysis. Automated underwriting systems have been here for a decade. The acceleration now is in the client-facing workflow, where AI handles intake, documentation, and preliminary approval.
The list continues through rank 25, covering roles in research assistance, financial analysis, HR screening, quality assurance testing, and basic coding. Each follows the same pattern: high cognitive repetition, low physical presence, clear input-output structure.
Where does your job land?
500+ occupations scored 0-10. Free lookup. Takes 60 seconds.
The Comparisons That Should Make You Uncomfortable
Same industry. Completely different futures. That's the part the summary statistics miss.
Surgeons score 3. Radiologists score 7. Both are physicians. Both spent a decade in training. One works with their hands in real-time, irreversible, physical space. The other interprets patterns in images. High score. Shared profession. Diverging trajectories.
Nurses score 2. Radiologists score 7. A nurse has more physical contact with patients in a single shift than a radiologist has in a month. The physical presence is the protection. The distance is the vulnerability.
The skills premium
AI skills now command a 56% salary premium in job postings. The people who learn to work with AI get paid more. The people who compete against it get paid less.
VP of Sales scores 6. SDRs under them score 8. As the bottom layer of a sales org automates, the structure above it compresses. You don't need three sales managers when AI handles top-of-funnel at scale. The second-order effect eats the middle, not just the bottom.
It's not about whether AI replaces your job. It's about whether someone using AI replaces you.
What the Score Tells You to Do Next
The score is not a verdict. It's a signal. Here's what to do with it.
-
Score 1-4: Protect your competitive advantage. Physical, relational, or highly contextual work is your moat. Don't let scope creep pull you into administrative tasks that AI will commoditize first.
-
Score 5-7: Identify the 30% of your tasks that are highest risk. Those are the tasks AI will absorb first. Either own those tasks more deeply (strategy, judgment) or build the adjacent skills that remain human after the transition.
-
Score 8-10: The window is 18-36 months. Not 10 years. Decide now whether you evolve the role, layer in AI skills to command the 56% premium, or pivot entirely. Waiting for certainty is the most expensive choice.
The global average score is 5.3. That means most workers are in the "restructure in 3-5 years" zone. Not eliminated. Transformed. The transformation gap, between those who adapt and those who don't, is where the real income divergence will happen.
The full survival analysis covers 12 specific actions across three risk tiers. This article covers the framework. The playbook goes deeper on the specific task-level shifts happening inside each of the 25 highest-risk roles right now.
Bottom Line
The jobs most at risk from AI in 2026 are not the ones at the bottom of the income ladder. They're the ones that look like careers but are actually just sequences of well-defined cognitive tasks. High salary. Respectable title. Hollow core.
The radiologist with 22 years of experience didn't lose his job to AI. He lost it to a hospital administrator looking at a spreadsheet, seeing two columns, and making a rational choice.
That's how this works. Not a dramatic disruption. A quiet, rational, cost-driven reallocation. One contract at a time.
The score tells you where you stand. What you do with it is the only variable left.
Find out where you stand
500+ occupations scored 0-10 on AI displacement risk. Free.
Keep Reading
Methodology: AI Displacement Scores are calculated using the JobHunter AI Displacement Index, which analyzes 500+ occupations across 12 risk factors including task automation potential, historical automation patterns, AI capability trajectories, and labor market dynamics. Data sources include Stanford's AI Index Report, Anthropic's capability research, Bureau of Labor Statistics employment projections, and O*NET task databases. Scores are updated quarterly. Learn more about our methodology.
Related AI Displacement Scores: Highest Risk · Medical Transcriptionists · Customer Service Representatives · Bookkeeping Accounting And Auditing Clerks