With the rise of AI commercialization, companies are launching AI products and services to consumers. However, consumers’ evaluation of these AI products and services vary significantly. Unlike previous studies that focused solely on the characteristics of either AI or the user, this study explores the causes and boundary conditions of this phenomenon more comprehensively by constructing a synthesized interaction model of "human-AI technology-task". Five scenario experiments revealed that novice users ( vs. expert users) preferred AI-dominated (vs. AI-assisted) products and services. The reason is that AI-dominated products tend to inspire positive (vs. negative) machine heuristics in novices (vs. experts). This elffect is amplified in contexts with high task complexity, while experts’ negative machine heuristic is weakened in low. These findings provide new insights and directions for understanding the distinctive responses of different users to various AI products, and also offer important guidance on product and service design and outreach strategy for companies.