AI and Economics in Marketing
(A2026-133548)

AUTHORS

Nicolas Padilla, London Business School; Hortense Fong, Columbia Business School; Shrabastee Banerjee, Tilburg University; Marco Kotschedoff, KU Leuven

ABSTRACT

AI and Economics in Marketing This Special Session brings together four cutting-edge talks that examine how generative AI and related computational advances are reshaping consumer behavior, firms decisions, and empirical modeling in marketing. Collectively, the papers demonstrate how AI both transforms markets and equips researchers with new tools to understand demand and measure these transformations. The session provides a unified perspective on how AI changes the economics of content, pricing, and engagement, while offering new methodological approaches for modeling consumers. The first talk, Generative AI Spurs Passive But Not Active Engagement with Content (Narang, Ahlbom, Banerjee), uses a large field experiment and complementary online studies to causally identify how AI-generated posts shape community participation. The authors show that AI assistance increases passive engagement (e.g., reactions) but not active engagement (e.g., comments), driven by a trade-off between informativeness and complexity. The second talk, Profit-Weighted Price Sampling for Structural Demand Estimation (Musalem, Padilla), introduces a new supply-side framework in which firms stochastically draw prices with probabilities increasing in profit. This model allows for price dispersion, multimodality, and multiple equilibria, offering a flexible alternative to standard first-order-condition approaches and providing a tractable basis for structural estimation. The third talk, The Impact of AI-Generated Product Line Extensions: Evidence from Machine-Narrated Audiobooks (Kotschedoff, Laitenberger), exploits Amazon’s introduction of synthetic-voice audiobooks to estimate cross-format spillovers. Using a large multi-country panel and dif-in-dif design, the authors show that AI-narrated audiobooks boost short-run eBook demand but cannibalize print demand in the long run. The fourth talk, Modeling Story Expectations to Understand Engagement: A Generative Framework Using LLMs (Fong, Gui, Yang), proposes a novel method for extracting expectations, uncertainty, and surprise from imagined story continuations generated by LLMs. Applying this framework to 30,000 story chapters, the authors show that belief-based features meaningfully enhance the explanation of narrative engagement. Together, these papers illuminate how generative AI acts simultaneously as an economic force, a technological shock, and an analytical tool. The session highlights the need to understand AI-mediated markets and provides researchers with frameworks to study emerging behaviors, model complexity, and design future measurement strategies.