AI's Labour Impact by Industry (September 2026)

The Macro Picture: Hybrid, Not Replacement

McKinsey's automation research (A Future That Works; Jobs Lost, Jobs Gained) found that less than 5% of occupations can be fully automated by currently available technology, while around 30% of the activities within most occupations can be. This task-level, not occupation-level, distinction is the key to understanding AI's labour impact. Economic forecasts diverge on the medium-term net effect: some models project net-positive employment over 5-10 years from productivity gains, while more recent modelling (including Goldman Sachs in 2026) points to a net drag on employment during the adoption transition, with displacement concentrated in routine task-heavy roles. The BLS 2026 workforce projections identify customer service representatives, data entry workers, and certain administrative roles as highest-displacement-risk; software developers, healthcare workers, and professional service roles as lower displacement risk with augmentation.

Customer Service: Partial Displacement, Measurable

Customer service is the clearest case of partial displacement in 2026. AI deflection rates of 40-60% at best-in-class deployments mean that 40-60% of tickets that previously required a human agent are now resolved without one. But: CS teams have not shrunk proportionally in most enterprises. Instead, deflection has absorbed volume growth, and the human agents handle higher-complexity work. The net effect in 2026 is lower headcount growth (not headcount reduction) in CS teams at companies with AI deployment. The most directly displaced role is the entry-level L1 CS representative handling FAQ-type tickets.

Sales: Mixed Signal on SDR Roles

The 22%/45%/2% data from Salesmotion (22% of sales teams fully replaced human SDRs, 45% run hybrid, and only 2% of full AI SDR deployments stick long-term) shows that AI SDR is a genuine displacement force in some companies but not across the board. The SDR role -- high-volume, templated outreach, cold prospecting -- is the highest-displacement-risk sales role. Account executive roles (relationship management, late-stage negotiation) are lower displacement risk. The net effect: SDR headcount is shrinking at companies that have adopted AI SDR tools; AE headcount is stable or growing. For individual SDRs, the career path is shifting toward AE earlier as AI handles the L1 prospecting work.

Engineering, Legal, Healthcare: Augmentation Dominant

In engineering: GitHub Copilot and similar tools are documented to increase developer productivity 40-55%; there is no documented evidence of engineering headcount reduction from AI tool adoption. The ROI is in doing more with the same team, not in reducing team size. In legal: AI contract review and research tools save hours per attorney per week; law firms have not reduced attorney headcount as a result; they have taken on more work per attorney. In healthcare: the largest controlled study of AI ambient scribes (JAMA, April 2026, ~1,800 adopters across five health systems) found documentation-time savings of roughly 16 minutes per eight patient-hours (more for heavy users), below the larger figures often cited in vendor marketing; physicians are not being replaced; physician shortages (not surpluses) remain the primary workforce challenge in healthcare. The augmentation pattern is clear in knowledge-work roles: AI increases output per professional; it does not reduce the number of professionals.

Calculator Reference

For per-role labour impact modelling, the aijobimpactcalculator.com calculator (sister site, Digital Signet AI cluster) provides a per-role displacement risk estimate based on task structure, AI substitutability, and regulatory constraints. Use it as a starting point for workforce planning conversations, not as a definitive forecast.

Frequently Asked Questions