■ Behind the Numbers ■

How We Score AI Replacement Risk

Every one of the 1131 jobs on this site carries a risk score from 0 to 100 — our estimate of how exposed that occupation is to replacement or radical transformation by AI and automation. Higher means more replaceable.

The scores are editorial estimates, not outputs of a single model. They are informed by the public body of automation research — including Oxford’s occupation-level automation-probability work (Frey & Osborne and its successors), the World Economic Forum’s Future of Jobs surveys, McKinsey’s task-automation analyses, and the U.S. Bureau of Labor Statistics’ occupational outlooks — combined with our own reading of what current AI systems can actually do. Where the research disagrees (it often does), we weigh task structure: how much of the job is routine information processing versus physical dexterity, human trust, and accountability.

We deliberately do not attach a precise citation to every number, because precise-sounding numbers are how this genre lies to you. The score is a considered judgment; the per-job pages explain the reasoning so you can disagree with it intelligently.

What the bands mean

76–100CRITICAL

The core tasks of the job are already automatable with current AI, and adoption is underway. Displacement is happening now or within this decade.

51–75HIGH

A majority of the job's tasks can be automated with current or near-term AI. Serious pressure on headcount by around 2030.

26–50MODERATE

AI transforms how the job is done — automating routine parts, amplifying the rest. The job changes substantially by 2040 but doesn't disappear.

0–25SAFE

The job leans on physical dexterity, human trust, emotional intelligence, or accountability that AI can't credibly replace in the foreseeable future.

What the score is not

It is not a prediction that you, personally, will lose your job — averages hide enormous variation in seniority, specialty, and employer. It is not career advice, and it is not a countdown clock. Jobs rarely vanish overnight; they erode task by task, hiring freeze by hiring freeze. The score tells you how much of your job’s task list AI is coming for, so you can move toward the parts it can’t do.

Scores are reviewed as the technology moves. When a frontier capability shifts the picture for a family of jobs — the way large language models did for writing-heavy work — we re-score the affected occupations.