Emerging AI technologies are transforming labor markets by altering job openings, skill requirements, and offered salaries-changes that may disproportionately affect vulnerable populations and job positions at high risk of automation. This study uses a dataset of LinkedIn job postings and AI-related patent abstracts to investigate how AI innovations relate to labor demand. We propose a Weighted AI Index (wAII), which measures job openings' exposure to AI by calculating the semantic similarity between job tasks extracted from job descriptions and relevant AI patents. Our analysis identifies industry sectors with high AI exposure-particularly those employing large shares of economically vulnerable workers. We also reveal thematic differences in job postings and show how higher exposure to current AI technologies may be associated with lower wages in some sectors. We validate the robustness of our findings through alternative configurations of task extraction and weighting schemes. Our results highlight AI's potentially uneven and multifaceted relationship with labor markets, underscoring implications for equitable AI and labor policy. We release our code and data to facilitate further research.

PDF