Abstract: In the contemporary digital economy, Artificial Intelligence (AI) recommendation systems have emerged as a pivotal tool for enhancing decision-making efficiency and driving sales growth. By leveraging big data and machine learning, these systems evoke a sense of “cognitive surprise” in user, where the algorithm seemingly understands the user better than they understand themselves. However, this high degree of personalization often triggers a “privacy-personalization paradox”, leading to algorithm aversion, privacy concerns, and “filter bubbles”. While existing research has predominantly focused on optimizing algorithmic accuracy or interface design, the linguistic dimension—specifically how promotional language can serve as a choice architecture to nudge consumer behavior—remains under-explored. This research introduces the innovative concept of “Promotional Text Nudging” to address the systemic challenges of AI recommendations, such as the cold-start problem and privacy-induced resistance. Drawing upon Nudge Theory and the research paradigm of conditional clauses in promotion decision-making, the study constructs a comprehensive “User-Product-Context-Language” analysis framework. It posits that promotional language, characterized by conditional restrictions and material incentives, can transform the rigid logic of algorithms into persuasive narratives that convey corporate benevolence and induce an emotional “hot state” in consumers.
The research is structured into four progressive sub-studies that track the transition from external linguistic stimuli to internal psychological drivers. First, the study investigates the completeness of promotional conditional clauses and its impact on recommendation attention. Utilizing Cognitive Resource Theory, it explores how the “If... then...” structure serves as a linguistic nudge. By explicitly stating the conditions for obtaining benefits, a complete conditional clause shifts the user’s limited cognitive resources from privacy risk assessment toward gain-oriented processing. This transition enhances promotional involvement and mitigates privacy concerns, ultimately maintaining sustained attention on the recommended content. Second, the research explores the self-relevance of promotional text and its role in reinforcing product preferences. This section examines how linguistic cues—such as the choice of grammatical subjects—induce an “illusion of fit”. In the low-context environment of AI recommendations, self-relevant language acts as a bridge, connecting the product’s objective attributes with the consumer’s subjective goals. This process of meaning construction transforms a digital recommendation into a personalized self-reward, particularly for hedonic products where emotional resonance is paramount. Third, the study focuses on the functional division of key promotional language regarding emotional arousal and intention guidance. Based on Self-Determination Theory, it argues that while AI recommendations often limit autonomy, a well-structured promotional choice framework allows users to internalize external incentives. Conditional language facilitates the perception of corporate benevolence and autonomy, while incentive language triggers prospective emotions such as hope and joy. Together, these mechanisms guide the user from mere recognition of the recommendation to an active intention to accept it. Fourth, the research adopts a holistic perspective on multimodal linguistic cues and contextual adaptation. It investigates how auxiliary promotional words, language-product-context alignment, and visual presentation (paralinguistics) synergistically drive instantaneous decisions. By applying Dual Coding Theory, the study reveals that the integration of vivid sensory language with optimized visual layouts enhances processing fluency. This heightened fluency fosters swift trust, allowing consumers to make intuitive, high-speed evaluations of the recommendation’s legitimacy and value. In conclusion, this research systematically elucidates the linguistic mechanisms through which promotional texts mitigate negative user responses and enhance AI recommendation effectiveness. Theoretically, it contributes to the intersection of linguistics and intelligent recommendation by integrating nudge theory with psycholinguistic principles to construct an “Attention-Sensemaking-Motivation-Decision” framework. This framework effectively bridges the gap between algorithmic logic and humanistic persuasion, offering a new path for linguistic empowerment in AI applications. Practically, the study provides a structured knowledge base for platforms to optimize recommendation texts, enabling them to move beyond mere “mind-reading” toward a more empathetic and effective “mind-winning” strategy. These insights are particularly valuable for resolving cold-start issues, weakening algorithm aversion, and fostering a more autonomous and positive digital consumption experience.

Key words: AI recommendation, promotional text, nudge theory, language perception, instantaneous decision- making