Barclays Unveils 3A Framework: Only 20% of Core Job Skills Highly Replicable by AI

Barclays published a research report on September 22 introducing the “AI Automation and Amplification Framework” (Barclays 3A Framework), a skill-based analytical model that maps 104 skills from the U.S. Department of Labor database to more than 830 occupations and assesses how cognitive and physical AI affect the labor market.

The report finds that only about 20% of core workplace skills are highly replicable by AI. Automation exposure is concentrated in a small set of occupations, with cognitive AI exposure highest in computer, engineering and science roles, and physical AI exposure concentrated in construction, production, installation, maintenance and repair. About 120 occupations show relatively low combined exposure, covering roughly 20% of U.S. workers.

Amplification effects, by contrast, spread more broadly. AI can take over administrative and document-processing tasks while enhancing human capabilities such as leadership, coaching and decision-making, with the highest cognitive amplification seen in roles like air traffic controllers and healthcare service managers.

Using job posting data, the report says the share of openings in occupations with the highest automation exposure fell from about 30% to around 25% since 2022. In the U.S., UK, France and Germany, the share of positions requiring AI skills rose from about 2% in 2019 to between 6% and 10% by August 2026. The share of AI-related postings in tech and data roles fell from above 70% to about 50%, while management roles increased from around 10% to about 25%, signaling that AI skill demand is spreading beyond technical functions.

The report also outlines a two-phase economic model for physical AI: an early phase in which humans act as “robot trainers” generating training data, followed by a shift toward codified “robot operation manuals” as a new form of intellectual property.