Abstract

As digital platforms increasingly rely on recommendation systems, understanding how these systems influence consumer behavior is essential. While AI-based recommendations offer strong personalization, research suggests that consumers may respond negatively when artificial intelligence is explicitly mentioned, particularly in emotional or pleasure-driven consumption contexts. This thesis examines how the framing of recommendations (AI-based, behavior-based, or similarity-based), product type (utilitarian or hedonic), and individual characteristics shape consumer responses.
Three studies were conducted. A pilot study with 184 participants assessed consumers’ familiarity with recommendation types and measured AI-related traits such as literacy, perceived capability, magical beliefs, fear, and usage. Study 1, with 490 participants, tested how recommendation framing and product type influenced trust, perceived relevance, and behavioral intentions. Study 2, with 361 participants, used a new shopping context and introduced novelty seeking as an additional moderator.
Results showed no consistent advantage for AI-based recommendations. However, consumers with higher perceived AI capability responded more positively to AI framing, while those with higher fear or lower AI usage were less receptive. Novelty-seeking individuals were more favorable toward similarity-based recommendations, particularly when evaluating hedonic products.
This research contributes to the literature on AI in marketing by showing that consumer responses to AI-based recommendations depend on both context and individual characteristics. Marketers should consider aligning recommendation framing with product category and customer traits to build trust and increase engagement.