Automation potential measures what fraction of a job's task portfolio is already within reach of current AI systems — not future theoretical capabilities, but tools available in 2026 such as large language models, computer vision systems, and robotic process automation.
A task is considered automatable when an AI system can complete it with equal or greater speed and quality compared to an average human practitioner, without requiring novel reasoning, physical presence, or real-time interpersonal judgment.
Automation potential is distinct from the AI Risk Score. A role might have 60% automation potential but still score only 40/100 on overall risk — because the remaining 40% of non-automatable tasks represent the highest-value, highest-paid work. Conversely, a role with 30% automation potential might score higher if the automatable tasks are core to its value proposition.
Researchers typically decompose a role into task categories (data processing, communication, physical manipulation, judgment) and assess each independently. The weighted average across task categories, adjusted for how central each task is to the role's output, produces the automation potential figure.