AI exposure is concentrated in thinking-heavy jobs, a study of 664 U.S. occupations finds

Study title: Occupational Vulnerability to AI-Driven Change: The Role of Skill Composition, Task Structure, and Psychosocial Buffers

Researchers analyzed 664 U.S. occupations across 128 skill dimensions, using K-means clustering to identify five distinct groups and scoring each for AI Exposure via AIOE and for protective job features via PBI, with PCA to examine how exposure and buffering relate to skill composition.

Read the study abstract

Background Automation risk is often framed as a question of which jobs can be automated. This perspective neglects how the structure of work itself shapes workers' vulnerability to AI-driven change. In occupational health research, job characteristics such as autonomy, task variety, and social interaction are known to influence worker well-being, yet are rarely incorporated into assessments of AI-related occupational risk. Methods To capture this complex vulnerability, this study distinguishes between Automation…

By the Truvace deskSeptember 7, 20261 min read1 sourceEvidence: Moderate · 60/100 — Observational study · n = 664 · 1 independent sourcePeer-reviewedTRV-2026-1004· cite & verify

Occupational Vulnerability to AI-Driven Change: The Role of Skill Composition, Task Structure, and Psychosocial Buffers — "Percentage of socially useless jobs by occupation (per workers' own perceptions)" by Authors of the study: Simon Walo is licensed under CC BY 4.0. To view a copy of this license, visit https://creativecommons.org/licenses/by/4.0/.

In brief

The pattern matters because it reframes automation risk from which jobs can be automated to how job structure shapes vulnerability, suggesting that autonomy, task variety, and social interaction can buffer high-exposure roles while low-exposure roles with weak buffers remain at risk, though the study does not test actual employment or health outcomes after AI adoption.

Main points

Study analyzed 664 U.S. occupations across 128 skill dimensions using K-means clustering to identify groups with similar skill profiles.
AI Exposure was quantified using the Artificial Intelligence Occupation Exposure (AIOE) score and protective features via Psychosocial Buffer Index (PBI).
Principal Component Analysis was used to examine relationships among AI Exposure, PBI, and occupational skill composition.

The problem

AI exposure concentrates in cognitively intense occupational clusters, creating multi-dimensional vulnerability that varies with job structure and buffering capacity.

The rundown

The authors used K-means clustering, an unsupervised machine learning algorithm, on 128 skill dimensions to group 664 occupations, then quantified AI Exposure with AIOE and protective features with PBI, and applied PCA to map their relationships.

Results showed five clusters with systematic differences in exposure and buffering, with high-exposure clusters varying in PBI and low-exposure roles still at risk for stress when buffering is limited, pointing to retraining and workflow redesign as interventions.