Scholarlink

<p>This research examines how recommender type (AI vs. human) interacts with recommendation set size (large vs. small) to shape consumers’ intention to choose from a recommendation set. Across three preregistered studies, a reversal effect emerges: Consumers are more likely to choose from a large (vs. small) recommendation set when the recommender is an AI agent, but from a small (vs. large) recommendation set when the recommender is a human agent (Study 1). This interaction is driven by perceived expertise. Specifically, larger (smaller) sets signal greater breadth (depth) of expertise for AI (human) recommenders, thereby increasing consumers’ likelihood of choosing from the recommendation set. Evidence for this mechanism is provided by mediation analyses (Study 2) and a moderation-of-process design (Study 3) that manipulates perceived benefits of knowledge-breadth vs. knowledge-depth. These findings offer actionable guidance for designing effective recommendation systems.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

When does size matter? Divergent effects of recommendation set size in AI versus human recommenders

  • 期刊论文

  • 发表日期：

    2026年5月28日

查看全文
  • Xiaohong Zhao,
  • Binglin Chen,
  • Zhiyong Yang

摘要

This research examines how recommender type (AI vs. human) interacts with recommendation set size (large vs. small) to shape consumers’ intention to choose from a recommendation set. Across three preregistered studies, a reversal effect emerges: Consumers are more likely to choose from a large (vs. small) recommendation set when the recommender is an AI agent, but from a small (vs. large) recommendation set when the recommender is a human agent (Study 1). This interaction is driven by perceived expertise. Specifically, larger (smaller) sets signal greater breadth (depth) of expertise for AI (human) recommenders, thereby increasing consumers’ likelihood of choosing from the recommendation set. Evidence for this mechanism is provided by mediation analyses (Study 2) and a moderation-of-process design (Study 3) that manipulates perceived benefits of knowledge-breadth vs. knowledge-depth. These findings offer actionable guidance for designing effective recommendation systems.