Methodology

How the AI job risk scores work

The model estimates job risk from tasks, not job titles alone. A job is more exposed when its core tasks are digital, repeatable, text-heavy, rules-based, or easy to verify.

Replacement score

Measures how much of the job could be automated or significantly reduced by AI systems, agents, robotics, RPA, or generative tools.

Augmentation score

Measures how much more productive a worker could become by using AI. High augmentation does not always mean high replacement.

Wage protection index

Estimates whether wages are protected by scarcity, licensing, trust, physical presence, regulation, safety, relationship value, or complex accountability.

Confidence score

Shows how confident the estimate is based on how clear the task pattern is. Roles with mixed responsibilities can have lower confidence.

Factors that increase AI risk

Repeatable digital work

Data entry, transcription, summaries, reporting, routine communication, and standard document production.

Low verification cost

Tasks where errors are easy to detect and correct are easier to automate safely.

High software adoption

Office, finance, marketing, legal, and technology roles are exposed because the work already happens in software.

Factors that protect jobs

Physical presence

Trades, care, field work, emergency response, and site-specific work are harder to automate fully.

Accountability

Licensing, liability, safety, and regulated sign-off slow full replacement even when AI assists the work.

Human trust

Care, persuasion, conflict handling, leadership, and relationship-based advice remain human-heavy.

Evidence signals considered

  • Task-level occupational analysis similar to O*NET work activities and ESCO skill groupings
  • Labour-market data from national statistics agencies such as BLS, ONS, ABS, and Statistics Canada
  • Automation and AI exposure research from organisations such as OECD, WEF, McKinsey, Goldman Sachs, and academic labour economists
  • Current capability patterns in LLMs, AI agents, computer vision, robotics, RPA, code generation, and clinical decision-support tools

How to read a score

A high replacement score does not mean every worker loses their job. It means a larger share of the role's tasks can be automated, compressed, or shifted to fewer workers. A high augmentation score means AI can make skilled workers much more productive.

Score bands

75-100: very high

Routine task exposure is substantial. Workers should move toward quality control, exceptions, client trust, or higher-skill adjacent work.

40-74: moderate to high

AI is likely to reshape the job. The role may remain viable, but the best workers will use AI as part of daily work.

0-39: lower risk

Replacement is harder, often because of physical presence, licensing, care, safety, or human accountability.

Important limitations

Scores are educational estimates, not predictions of guaranteed job loss. Local wages, regulation, employer adoption, unionization, and economic cycles can change the result. The site should be updated as AI capabilities and labour-market evidence change.