워킹 페이퍼Working Paper AI & Labor SSCI

Comparing Occupational AI Exposure Measures with Platform-Derived Usage Traces: Quantity, Mode, and Sample Selection

Kim, Y. (Co-Author)

Social Science Computer Review (Under Review) (2026)

영향력 지수Impact Factor: 4.1
분위Quartile: Q1

초록Abstract

Occupational AI exposure indices are widely used to forecast labor-market impacts, yet they are constructed from task descriptions rather than observed behavior. This paper compares theory-based exposure measures with platform-derived AI usage traces, showing where the two diverge in quantity of use, mode of use (automation versus augmentation), and the sample selection embedded in platform data — with implications for how exposure scores should be interpreted in policy analysis.

키워드Keywords

AI Exposure
Occupations
Usage Traces
Automation
Augmentation
Sample Selection
Labor Market

연구 방법Methodology

연구 분야Research Areas: AI & Labor · Computational Social Science
방법론Methodology: Measurement Comparison · Platform Data Analysis
통계 기법Statistical Methods: Correlation Analysis · Selection Diagnostics
분석 도구Software: R · Python

Overview

Exposure indices tell us which occupations could be transformed by AI; platform usage traces tell us which ones actually use it, and how. Matching the two at the occupation level, this study quantifies the gap between predicted exposure and observed usage, separates automation-type from augmentation-type use, and characterizes the sample selection that platform-derived measures inherit.

Status

Current Status: Under Review (submitted August 2026) Journal: Social Science Computer Review (SSCI, Q1) Author Role: Co-author