AI Resilience Classification Methodology
How we assess careers for AI resilience using a transparent, additive scoring system based on three key dimensions.
The Four Tiers
Every career receives one of four classifications based on their total score.
| Tier | Description |
|---|---|
| AI-Resilient | Strong human advantage or growing demand protects this career from AI displacement |
| AI-Augmented | AI assists this work but human skills remain essential |
| In Transition | This career is being transformed by AI; adaptation and skill evolution needed |
| High Disruption Risk | High AI exposure combined with declining demand creates significant risk |
How We Calculate It
We use a simple additive scoring system. Each of three dimensions earns 0-2 points, for a maximum total of 6 points. The total determines the classification.
Scoring Breakdown
| Dimension | What We Measure | Points |
|---|---|---|
| AI Exposure | % of tasks AI can accelerate by 50% | 0-2 |
| Job Growth | BLS 2024-2034 employment projections | 0-2 |
| Human Advantage | EPOCH framework score (1-25) | 0-2 |
Total Score → Classification
| Score | Classification |
|---|---|
| 5-6 | AI-Resilient |
| 3-4 | AI-Augmented |
| 2 | In Transition |
| 0-1 | High Disruption Risk |
The Three Dimensions Explained
Each dimension is scored independently based on objective data sources.
What it measures
The percentage of an occupation's tasks where AI tools (specifically Large Language Models like ChatGPT) could reduce completion time by at least 50% while maintaining quality.
How we calculate it
We use the "GPTs are GPTs" research (Eloundou et al., 2023) from OpenAI and University of Pennsylvania, which analyzed task-level LLM exposure.
| AI Exposure | Meaning | Points |
|---|---|---|
| Low | <25% of tasks exposed | +2 |
| Medium | 25-50% of tasks exposed | +1 |
| High | >50% of tasks exposed | +0 |
Why lower exposure = more points
Careers with fewer AI-exposed tasks are more protected from disruption. A low AI exposure score means most of what you do in this job cannot be easily done or accelerated by AI.
Example: Electrician
| Dimension | Assessment | Points |
|---|---|---|
| AI Exposure | Low — Only ~4% of tasks can be accelerated by AI. The work is physical. | +2 |
| Job Growth | Growing — 6% projected growth. Infrastructure investment drives demand. | +2 |
| Human Advantage | Strong — High presence, high judgment, skilled hands-on work. | +2 |
| Total | 6 → AI-Resilient |
Limitations & Disclaimers
- 1. AI capabilities are evolving rapidly. The AI exposure data is based on 2023 research. New AI capabilities could change exposure levels for some occupations.
- 2. Individual jobs vary significantly. This classification represents aggregate trends for occupation categories. Your specific job depends on employer, location, specialization, and how you adapt.
- 3. Projections are estimates, not guarantees. BLS projections may not account for major economic shifts, policy changes, or breakthroughs.
- 4. EPOCH scores involve judgment. While grounded in O*NET data, EPOCH scoring requires qualitative assessment.
- 5. This is not career advice. This tool informs your thinking but should not make decisions for you.
Data Sources
| Source | What we use it for | Year |
|---|---|---|
| GPTs are GPTs (OpenAI/UPenn) | AI Exposure (primary) | 2023 |
| AIOE Dataset (Felten et al.) | AI Exposure (fallback) | 2021 |
| BLS Employment Projections | Job growth outlook | 2024-2034 |
| O*NET 30.1 | Task descriptions, work activities | 2024 |
| EPOCH Framework | Human advantage scoring | 2024 |
Questions or Feedback?
We're committed to making this methodology as accurate and useful as possible. If you have expertise in a career we've classified and think we got it wrong, please let us know.