The effect of error acknowledgment on consumer responses to human and AI influencers

Abstract

Human and artificial intelligence (AI) influencers increasingly share product recommendations, yet how consumers react when these actors respond to an informational error remains unclear. This study examines whether acknowledging responsibility for an error, rather than denying it, produces different consumer responses for human and AI influencers. A 2 (influencer type: human vs. AI) × 2 (error response: acknowledgment vs. denial) between-subjects scenario experiment was conducted with 496 social media users. Data were analysed using MANOVA, ANOVAs, and moderated serial mediation with 5,000 bootstrap samples. Error acknowledgment increased perceived warmth, parasocial interaction, trust, and purchase intention for both influencer types, although effects were substantially stronger for human influencers. A crossover interaction emerged for perceived competence: acknowledgment raised competence for the human influencer but lowered it for the AI influencer. Indirect effects on trust and purchase intention operated mainly through warmth, whereas the competence penalty reached behavioural intentions only serially via parasocial interaction. Findings specify boundary conditions of the pratfall effect and reveal a competence cost of AI transparency.