AIS Insights AI Podcast
Paper Number
ICIS2026-3257
Paper Type
Short
Abstract
Large language models (LLMs) are emerging as new shopping gatekeepers by mediating consumer search, product discovery, and recommendations. We study when marketplace platforms should cooperate with external LLMs and how such cooperation reshapes seller competition. We develop a stylized game-theoretic model in which cooperation introduces an LLM-mediated recommendation channel and sellers can engage in Generative Engine Optimization (GEO). Our analysis shows that LLM-induced traffic may intensify price and GEO competition and reduce marketplace advertising. In contrast, higher LLM inference accuracy softens GEO-based competition and can complement platform advertising. Cooperation is profitable when the LLM provides sufficient traffic scale and inference accuracy. Moreover, sellers may be worse off under cooperation when the LLM places high weight on GEO, leading to a rent-dissipating GEO arms race. Overall, our study innovatively incorporates sellers’ GEO behaviors and provides several theoretical insights into LLM-mediated shopping.
Recommended Citation
Fan, Bingbing and Zhang, Xin, "AI as Shopping Gatekeepers: Marketplace Cooperation with External LLMs" (2026). ICIS 2026 Proceedings. 12.
https://aisel.aisnet.org/icis2026/econ/econ/12
AI as Shopping Gatekeepers: Marketplace Cooperation with External LLMs
Large language models (LLMs) are emerging as new shopping gatekeepers by mediating consumer search, product discovery, and recommendations. We study when marketplace platforms should cooperate with external LLMs and how such cooperation reshapes seller competition. We develop a stylized game-theoretic model in which cooperation introduces an LLM-mediated recommendation channel and sellers can engage in Generative Engine Optimization (GEO). Our analysis shows that LLM-induced traffic may intensify price and GEO competition and reduce marketplace advertising. In contrast, higher LLM inference accuracy softens GEO-based competition and can complement platform advertising. Cooperation is profitable when the LLM provides sufficient traffic scale and inference accuracy. Moreover, sellers may be worse off under cooperation when the LLM places high weight on GEO, leading to a rent-dissipating GEO arms race. Overall, our study innovatively incorporates sellers’ GEO behaviors and provides several theoretical insights into LLM-mediated shopping.
When commenting on articles, please be friendly, welcoming, respectful and abide by the AIS eLibrary Discussion Thread Code of Conduct posted here.
Comments
20_Economics_of_commerce