A detailed look at our methodology for calculating AI displacement risk scores. Transparency about our data sources, model, and limitations.
Transparency is core to our mission at displacement.ai. This article explains exactly how we calculate AI displacement risk scores, the data sources we rely on, and the limitations of our methodology. We believe you should understand how we arrive at our conclusions so you can make informed decisions about your career.
As of this writing, we have analyzed 1000 occupations, with an average risk score of 71%. Of these, 790 fall into the critical risk category, while many others show strong resistance to automation.
The International Monetary Fund estimates that 40% of global employment is exposed to AI, with this figure rising to 60% in advanced economies. The OECD finds that 27% of jobs in member countries are at high risk of automation. These aren't distant predictions—they describe changes happening now.
Yet most workers lack the data to assess their personal risk. Our goal is to provide that information: specific, actionable intelligence about how AI will affect individual careers.
Our Displacement Index assigns each job a risk score from 0-100, where higher scores indicate greater vulnerability to AI automation. The score is calculated using four weighted factors, each informed by peer-reviewed research on technological unemployment.
Analysis of individual job tasks and their susceptibility to current AI capabilities, including LLMs, computer vision, and robotics.
Projected advancement of AI systems relevant to the job within 5-10 years, based on current research and industry adoption patterns.
Cost-benefit analysis considering wages, industry profit margins, and historical automation investment patterns.
Regulatory requirements, licensing, physical constraints, and social acceptance factors that protect against automation.
The foundation of our approach is task-level analysis. We don't ask "Will this job be automated?" but rather "Which specific tasks within this job can AI perform, and how well?"
For each occupation, we decompose the job into its component tasks using O*NET database classifications and Bureau of Labor Statistics occupational data. Each task is then evaluated against five categories of AI capability:
Understanding where AI is today—and where it's heading—is essential for accurate risk assessment. We continuously monitor developments across domains:
Models like GPT-4, Claude, and Gemini demonstrate strong performance in text generation, analysis, coding, and reasoning tasks. The Stanford AI Index reports that frontier models now match or exceed human performance on many benchmarks.
Image and video analysis systems have reached production quality for quality inspection, medical imaging, surveillance, and autonomous navigation.
Physical automation advances more slowly than cognitive AI. While warehouse automation is mature, tasks requiring adaptation to unpredictable environments remain challenging.
Domain-specific AI for medical diagnosis, legal research, financial analysis, and drug discovery is advancing rapidly with significant industry investment.
Our analysis synthesizes data from multiple authoritative sources:
Our methodology draws from 68+ peer-reviewed reports from these institutions. See our full methodology page for the complete list of citations.
We classify risk scores into four tiers, each with different implications for career planning:
Strong human advantage. Limited automation expected within 10 years. Natural barriers protect these roles.
Some tasks automatable. Role likely to evolve significantly. Hybrid human-AI work expected.
Majority of tasks automatable. Significant workforce reduction expected within 5-7 years. Proactive skill development advised.
Most tasks highly automatable. Substantial displacement likely within 3-5 years. Immediate career transition planning recommended.
Our methodology has important limitations that users must understand when making career decisions:
We update our risk scores monthly as AI capabilities evolve and new data becomes available. We track real-world employment trends to validate and refine our methodology over time. When we're wrong, we publish corrections.
Our goal is not to cause panic or provide false reassurance, but to give workers the information they need to make informed decisions about their careers. We believe everyone deserves access to this data, which is why our basic analysis is free.
For complete documentation including all citations, data sources, and technical details, see our comprehensive methodology page.
Full Methodology DocumentationA risk score is a starting point, not a verdict. Here's how to use our analysis productively:
If you use our data or methodology in research, please cite:
displacement.ai. (2026). AI Job Displacement Risk Index: Methodology. https://displacement.ai/methodology
Help others understand AI's impact on careers.
Get the latest AI job displacement insights, risk score updates, and career recommendations delivered to your inbox every week.