📐 How We Calculate AI Risk

Our risk scores are derived from a multi-factor model combining automation research, occupational data, and AI capability assessments. Here's exactly how it works.

The 7 Scoring Factors

Each career is evaluated across seven dimensions, each scored 0–10:

Factor0 = …10 = …Effect on Risk
Routine TasksFluid, unpredictableHighly scripted & repetitive↑ Increases risk
Data ProcessingNo data workPrimary role is data analysis↑ Increases risk
CreativityNo original thinkingConstantly generates novel ideas↓ Reduces risk
Social SkillsNo human interactionDeep empathy & relationship-building↓ Reduces risk
Physical DexterityPurely mental workComplex unstructured physical tasks↓ Reduces risk (short-term)
Decision MakingFollows strict rulesHigh-stakes judgment under uncertainty↓ Reduces risk
AdaptabilityDomain is staticRequires constant learning & flexibility↓ Reduces risk

The Formula

raw_score = (Routine × 1.8) + (Data × 1.2)
− (Creativity × 1.5) − (Social × 1.0)
− (Physical × 0.8) − (Decision × 1.0)
− (Adaptability × 0.7)

current_risk% = normalize(raw_score, min=−50, max=30) × 100
5-year_risk% = current_risk × 1.25 (capped at 98%)
10-year_risk% = current_risk × 1.55 (capped at 99%)

The weights reflect current AI strengths: language models and data tools are far ahead of robotics, physical manipulation, and genuine creative reasoning.

Risk Tiers

TierScore RangeMeaning
🟢 Very Low0–19%Strong human-centric elements. AI will augment, not replace.
🔵 Low20–39%Some tasks will be automated. Role evolves but survives.
🟡 Moderate40–59%Significant disruption ahead. Reskilling recommended.
🔴 High60–74%Core tasks are automatable. Role demand will decline.
🟣 Critical75–100%Most tasks replaceable today. Transition planning urgent.

Known Limitations

Geography matters: AI adoption varies widely by country, company, and sector.

Specialization matters: A general accountant faces higher risk than a forensic accountant or CFO.

Regulatory lag: Even technically automatable roles (e.g. radiologist) may be protected by regulation for years.

Unknown careers: For careers not in our database, we infer factors from keywords — accuracy is lower for niche roles.

AI capabilities evolve: Physical robotics is improving rapidly. Physical tasks will face more risk after 2027.

References & Research

Our model is inspired by:

• Frey & Osborne (2013) — The Future of Employment (Oxford)

• McKinsey Global Institute (2017, 2023) — Jobs Lost, Jobs Gained

• World Economic Forum — Future of Jobs Report 2023

• OpenAI / Anthropic capability benchmarks (2024–2025)

← Analyze a Career