The AI Displacement Index

Telemarketer

99%

The most at-risk occupation in the Oxford dataset

Of the 758 occupations analyzed by Oxford researchers Frey and Osborne, the telemarketer sits at the very top of the automation risk index — a 99% probability of computerization. The reasons are structural: the role involves scripted communication, high repetition, and clear performance metrics that make it ideally suited to AI replacement.

The transformation is already underway. AI-powered voice agents can now conduct outbound sales calls indistinguishable from human callers in controlled tests. Companies including Salesforce, HubSpot, and dozens of AI startups now offer automated outreach tools that operate at a fraction of the cost of human teams.

The numbers are stark. The US Bureau of Labor Statistics projected a 28% decline in telemarketer employment between 2020 and 2030 — a projection made before the current generation of large language models existed. The actual decline is likely to significantly exceed this figure.

What does this mean for the estimated 400,000 people employed as telemarketers in the US alone? The honest answer is that direct replacement is coming, and the timeline is short. The skills that define the role — scripted persuasion, objection handling, product knowledge — are precisely the capabilities that current AI systems handle best.

The adjacent opportunity lies in AI oversight: managing, training, and quality-controlling automated outreach systems. This requires understanding of both sales strategy and AI capabilities — a combination that commands significantly higher compensation than the roles being displaced.

Occupational Therapist

0.35%

Among the most automation-resistant occupations in the dataset

At the opposite end of the spectrum sits the occupational therapist — a role so deeply embedded in human physical presence, clinical judgment, and therapeutic relationship that Oxford researchers assigned it a 0.35% automation probability. This is not an oversight. It reflects the genuine difficulty of replicating what occupational therapists actually do.

The role requires simultaneous assessment of physical capability, psychological state, environmental factors, and long-term rehabilitation goals — all while maintaining a therapeutic relationship with a patient who may be frightened, in pain, or cognitively impaired. No current AI system manages this combination with anything approaching human competence.

Demand is structural and growing. Aging populations across developed economies are increasing the need for rehabilitation services. The US Bureau of Labor Statistics projects 12% growth in occupational therapy employment through 2032 — significantly faster than average. This is a profession gaining relevance as AI displaces other work.

For workers in high-risk occupations considering a transition, healthcare rehabilitation careers represent one of the most defensible long-term positions in the labor market. The barrier to entry — graduate education and clinical training — is also the primary protection against displacement.

The data on AI displacement is simultaneously alarming and reassuring, depending on which numbers you choose to highlight. Both readings are defensible. Both are incomplete.

The alarming reading: hundreds of millions of jobs face genuine disruption within a decade. The occupations most at risk are concentrated among the people least equipped to manage rapid career transitions — lower-income workers, older employees, those in regions with limited retraining infrastructure.

The reassuring reading: every previous wave of automation created more jobs than it destroyed. The calculator did not eliminate accountants. The ATM did not eliminate bank tellers. In both cases, the role evolved and total employment grew. There is reasonable historical evidence to expect the same from AI.

The honest reading: we do not know. The current wave of AI is qualitatively different from previous automation. Earlier tools automated physical or narrowly defined cognitive tasks. Large language models automate general reasoning and communication — the capabilities that previously made humans irreplaceable in white-collar work.

What we can say with confidence: the transition will not be evenly distributed. Workers who understand AI capabilities and adapt their skills accordingly will fare significantly better than those who do not. The gap between these groups is likely to widen significantly over the next decade.

This is why this report exists. Not to generate fear, but to provide clarity. Understanding your risk is the first step toward managing it.

Volume II · April 2026 — Coming soon

Volume III · May 2026 — Coming soon