To reveal or conceal: AI identity disclosure strategies for merchants

Abstract

Online merchants face a dilemma in deciding whether to disclose their use of AI-driven services, which offer cost-effective, human-like interactions but may trigger customer resistance from those who prefer human engagement. The study employs a game-theoretic model to examine AI identity disclosure strategies, incorporating consumer preferences for algorithms and their ability to detect undisclosed AI usage. Findings indicate that nondisclosure is ineffective in markets with a high proportion of sophisticated consumers who can recognize AI involvement. However, in markets dominated by naive consumers, where AI identity is less detectable, disclosure remains the preferred strategy, particularly when AI service quality is perceived as low, to mitigate potential backlash. Comparing pure AI and human-AI collaboration service modes, merchants derive greater benefits from nondisclosure in collaboration settings. Furthermore, greater consumer aversion to AI does not necessarily push merchants toward nondisclosure, as its effectiveness depends on service quality rather than aversion alone. This study also highlights the dual role of AI anthropomorphism: it can increase consumer acceptance of nondisclosure while simultaneously making disclosure more appealing. Under nondisclosure, improvements in service quality generally enhance consumer surplus, but sophisticated consumers may gain less surplus than naive consumers. Lastly, mandating AI identity transparency does not always maximize social welfare, challenging the assumption that compulsory disclosure is universally beneficial. These findings offer valuable insights into human-AI collaboration and inform the design of algorithm transparency policies in AI-driven services.