Scholarlink

<p>Anthropomorphism in human-AI interaction is a critical yet paradoxical phenomenon shaping user engagement with Large Language Models (LLMs) and AI chatbots. Despite a surge in scholarly interest, the field lacks a systematic synthesis that uncovers the interdependencies between the theories, contexts, characteristics, and methods defining this research landscape. This review addresses this critical gap by employing the theory-context-characteristic-method (TCCM) framework to critically analyze and integrate 57 articles published between 2010 and 2025. Following PRISMA guidelines and quality assessment via the Critical Appraisal Skills Programme (CASP), our analysis reveals that while research in this domain is predominantly grounded in theories such as computers as social actors (CASA), mind perception theory, and the technology acceptance model, the efficacy of anthropomorphism is fundamentally contingent upon context. Studies delineate interaction stakes into high-stakes and low-stakes contexts, primarily involving Western and East Asian user groups, revealing a significant cultural bias that distorts theoretical constructs such as mind perception and trust, particularly when applied to Global South contexts. Key characteristics shaping engagement range from perceived humanness and trust to under-researched negative effects such as psychological reactance and AI anxiety, with recent evidence showing that anthropomorphic features such as empathy can increase user anxiety. The methodological landscape, dominated by controlled experiments and survey research, has established initial causal links but reveals significant limitations due to over-reliance on self-reported data. We highlight how emerging work using neurophysiological and cognitive models, such as EEG and eye-tracking, is beginning to address this gap. This first comprehensive TCCM-based synthesis of anthropomorphism in user interaction with LLMs and AI chatbots offers an integrative framework, outlining critical research gaps and a targeted future agenda for scholars, designers, and policymakers.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A TCCM-Based Systematic Literature Review of Anthropomorphism in Human-AI Interaction and Future Research Agenda

  • 期刊论文

  • 发表日期：

    2026年1月20日

查看全文
  • Bashir Ansari,
  • Ahmed Sameer

摘要

Anthropomorphism in human-AI interaction is a critical yet paradoxical phenomenon shaping user engagement with Large Language Models (LLMs) and AI chatbots. Despite a surge in scholarly interest, the field lacks a systematic synthesis that uncovers the interdependencies between the theories, contexts, characteristics, and methods defining this research landscape. This review addresses this critical gap by employing the theory-context-characteristic-method (TCCM) framework to critically analyze and integrate 57 articles published between 2010 and 2025. Following PRISMA guidelines and quality assessment via the Critical Appraisal Skills Programme (CASP), our analysis reveals that while research in this domain is predominantly grounded in theories such as computers as social actors (CASA), mind perception theory, and the technology acceptance model, the efficacy of anthropomorphism is fundamentally contingent upon context. Studies delineate interaction stakes into high-stakes and low-stakes contexts, primarily involving Western and East Asian user groups, revealing a significant cultural bias that distorts theoretical constructs such as mind perception and trust, particularly when applied to Global South contexts. Key characteristics shaping engagement range from perceived humanness and trust to under-researched negative effects such as psychological reactance and AI anxiety, with recent evidence showing that anthropomorphic features such as empathy can increase user anxiety. The methodological landscape, dominated by controlled experiments and survey research, has established initial causal links but reveals significant limitations due to over-reliance on self-reported data. We highlight how emerging work using neurophysiological and cognitive models, such as EEG and eye-tracking, is beginning to address this gap. This first comprehensive TCCM-based synthesis of anthropomorphism in user interaction with LLMs and AI chatbots offers an integrative framework, outlining critical research gaps and a targeted future agenda for scholars, designers, and policymakers.