Brand recommendations by generative AI chatbots: A comparative analysis in consumer decision-making
This study explored the algorithmic gatekeeping role of generative AI chatbots in consumer decision-making. Using algorithmic content analysis and an audit design, it examined the brand recommendations of the three most widely used chatbots in Türkiye: ChatGPT, Gemini and Copilot. Four product categories were selected by crossing involvement level and consumption motivation: detergent, chocolate, smartphone and an international holiday. Each prompt was submitted once to each model in a neutral and an emotional version, and the resulting 24 responses were coded descriptively. Neutral prompts mostly elicited technical and functional arguments, whereas emotional wording shifted the persuasion frame toward emotional appeals in 8 of 12 responses; predominantly emotional responses appeared only in hedonic categories. In the holiday category, all three models framed even the neutral prompt in romantic terms. The models typically offered three to ten brands rather than a single one, yet ChatGPT stated a first-person preference in every response and displayed priced shopping cards in five of eight. The findings suggest that generative AI can act as an intermediary that shapes brand visibility and has the potential to steer consumer preferences, raising questions for marketing ethics and large language model optimization.
Authors
- Erdem Varol (ORCID: https://orcid.org/0000-0003-3940-2122)
Institutions
- Marmara University (TR)
Publication Details
- Journal
- Opus uluslararası toplum araştırmaları dergisi
- Published
- 2026-11-19
- DOI
- https://doi.org/10.26466/opusjsr.2057877
- Primary Topic
- AI in Service Interactions
- Type
- article
- Field-Weighted Citation Impact
- 0.00