Home Articles Abstract
Research Article

Productivity Spillovers from Generative AI: How ChatGPT Adoption Reshapes Digital Engagement

Haeun Kim1 · Sungho Park1 · Juwon Hong1 · Dahye Lee1

1 Seoul National University

Published: January 2025 · Vol. 54 No. 6 · pp. 1939-1964

DOI: https://doi.org/10.17287/kmr.2025.54.6.1939

Full Text

Abstract

This study demonstrates how generative AI, specifically ChatGPT, influences user engagement across digital platforms beyond direct substitution or functional overlap. Using detailed user-level data from Nielsen Korea’s KoreanClick+ panel and a Difference-in-Differences design with Propensity Score Matching, we show that ChatGPT adoption is associated with increased total device usage, with stronger effects observed on PC devices and among users with lighter baseline engagement. Furthermore, users primarily reallocate engagement toward platforms they already frequent, suggesting that generative AI reinforces existing usage patterns rather than leading to the expansion into new platforms. We propose that these indirect behavioral spillovers may be driven, in part, by productivity gains that reduce cognitive effort, a finding consistent with Cognitive Load Theory. These findings highlight the broader impact of generative AI on digital engagement and provide a potential theoretical lens for understanding the underlying mechanisms.
Keywords: Generative AIChatGPTDigital EngagementCognitive LoadSpillover