Artificial intelligence (AI) has profound implications for both communication and persuasion. We consider how AI complicates and promotes rethinking of persuasion theory and research. We define AI-based persuasion as a symbolic process in which a communicative-AI entity generates, augments, or modifies a message—designed to convince people to shape, reinforce, or change their responses—that is transmitted to human receivers. We review theoretical perspectives useful for studying AI-based persuasion—the Computers Are Social Actors (CASA) paradigm, the Modality, Agency, Interactivity, and Navigability (MAIN) model, and the heuristic-systematic model of persuasion—to explicate how differences in AI complicate persuasion in two ways. First, thin AI exhibits few (if any) machinic (i.e., AI) cues, social cues might be available, and communication is limited and indirect. Second, thick AI exhibits ample machinic and social cues, AI presence is obvious, and communication is direct and interactive. We suggest avenues for future research in each case.

中文翻译：

---

人工智能（AI）时代的说服：基于人工智能的说服的理论和复杂性

人工智能 (AI) 对沟通和说服都有深远的影响。我们考虑人工智能如何使说服理论和研究变得复杂并促进重新思考。我们将基于 AI 的说服定义为一个象征性的过程，在该过程中，交流性 AI 实体生成、增强或修改一条消息——旨在说服人们塑造、强化或改变他们的反应——并将其传输给人类接收者。我们回顾了对研究基于 AI 的说服有用的理论观点——计算机是社会参与者 (CASA) 范式、模态、代理、交互性和可导航性 (MAIN) 模型，以及说服的启发式系统模型——以阐明在人工智能以两种方式使说服复杂化。首先，瘦人工智能表现出很少（如果有的话）机械（即人工智能）线索，社交线索可能是可用的，沟通是有限的和间接的。第二，厚AI表现出丰富的机械和社交线索，AI存在明显，交流直接互动。我们建议在每种情况下进行未来研究的途径。