AI & careers
How much AI is likely to change the daily work in each career.
Changing fast does not mean disappearing. Careers in the Transforming band are being reinvented, not removed.
What keeps it steady 0 to 3
SteadyEvolvingTransformingHands-onIn-person careLicensed & accountable
Steady · 6 careers: the daily work is likely to stay much as it is
Dentistry Steady Hands-onCareLicensed Physiotherapy Steady Hands-onCareLicensed Veterinary Science Steady Hands-onCareLicensed Nursing Steady Hands-onCareLicensed Occupational Therapy Steady Hands-onCareLicensed Medicine Steady Hands-onCareLicensedEvolving · 7 careers: parts of the daily work will change
Social Work Evolving Hands-onCareLicensed Pharmacy Evolving Hands-onCareLicensed Teaching Evolving Hands-onCareLicensed Engineering Evolving Hands-onCareLicensed Architecture Evolving Hands-onCareLicensed Psychology Evolving Hands-onCareLicensed Law Evolving Hands-onCareLicensedTransforming · 10 careers: the daily work is being reinvented, and new kinds of roles are appearing
Agricultural Science Transforming Hands-onCareLicensed Natural Sciences Transforming Hands-onCareLicensed Quantity Surveying Transforming Hands-onCareLicensed Chartered Accountancy Transforming Hands-onCareLicensed Actuarial Science Transforming Hands-onCareLicensed Finance & Economics Transforming Hands-onCareLicensed Business Management Transforming Hands-onCareLicensed Computer Science & IT Transforming Hands-onCareLicensed Humanities (BA) Transforming Hands-onCareLicensed Data Science & Statistics Transforming Hands-onCareLicensedThe line runs from changes little (left) to changes a lot (right). Dots, 0 to 3: how much hands-on work, in-person care and a licence protect the work. Open a career for its reasoning.
How the score works
- The bands: behind each dot is a robustness score from 0 to 100 (the dot sits at 100 minus the score). Steady is 60 or more, Evolving 35 to 59, Transforming under 35: the left, middle and right of the line. AI robustness (0-100) = 50% published task-exposure research (inverted) + 50% three protective factors scored 0-3 by MatricMap: hands-on physical work, in-person human care and trust, and legal accountability (a licensed person must sign off). Exposure is the average of each occupation’s percentile among all US occupations in two peer-reviewed datasets.
- Task exposure: the share of a job's tasks that AI tools can already help with. High exposure means the way the work is done will change, not that the job goes away.
- Reinvented, not removed: routine tasks shrink, people who learn to direct AI tools become far more productive, and new kinds of roles appear. Steady careers are not untouched either: AI is already taking over paperwork in medicine and nursing.
- The dots are MatricMap's own judgement, scored 0 to 3; each career's reasoning shows when you hover over its row.
- Sources: the exposure figures come from published research on the United States labour market, used as the best available proxy for the same professions elsewhere. The bands are MatricMap's estimate, not a forecast of jobs or pay, and nobody can predict this with certainty.
- Coverage: 23 broad fields; new specialist guides are not given their parent field's score. Browse all guides.
- Eloundou, Manning, Mishkin & Rock (2024), "GPTs are GPTs", Science 384 - occupation-level exposure (human-rated beta)
- Dataset: openai/GPTs-are-GPTs (occ_level.csv)
- Felten, Raj & Seamans (2023), "How will Language Modelers like ChatGPT Affect Occupations and Industries?" - language-modelling AIOE
- Dataset: AIOE-Data/AIOE