EXPOSURE TO AI
35%
MODERATE
OBSERVED IN REAL USE · Anthropic 2026
4%
of this role’s work is already showing up in real Claude usage (Anthropic Economic Index).
The boring parts are leaving. The judgment stays.
35% of this role’s O*NET tasks are within reach of today’s AI. That is the core-weighted exposure score from Eloundou et al. 2023 (“GPTs are GPTs”). It measures a capability ceiling, not a headcount forecast. In the blast radius. A real slice of the work is already automatable. The rest isn’t.
WHAT AI CAN ALREADY DO
- Maintain records of evidence and write and review reports
- Submit evidence to supervisors, crime labs, or court officials for legal proceedings
- Photograph crime or accident scenes for evidence records
- Create sketches and diagrams, by hand or computer software, to depict crime scenes
- Serve as technical advisor and coordinate with other law enforcement workers or legal personnel to exchange information on crime scene collection activities
WHAT IT STILL CAN’T
- Testify in court and present evidence
- Dust selected areas of crime scene and lift latent fingerprints, adhering to proper preservation procedures
- Package, store and retrieve evidence
- Coordinate or conduct instructional classes or in-services, such as citizen police academy classes and crime scene training for other officers
- Perform emergency work during off-hours
THE HONEST PART. A percentage is not a pink slip. High exposure usually means a role shrinks and shifts toward judgment, direction and responsibility: the parts a model can’t sign its name to. Exposure ≠ displacement. Breathe.
"My job is 35% cooked. What’s yours?"
SOURCES: O*NET 30.3 occupational tasks · Eloundou et al. 2023 (“GPTs are GPTs”,
arXiv:2303.10130) · Anthropic Economic Index 2026 (CC-BY) |
how this is calculated |
last updated 2026-07-16