Scholarlink

<p>The evolution of artificial intelligence (AI) is currently undergoing a&#xa0;paradigm shift from reactive chatbots to autonomous AI agents that are capable of independently planning and executing complex sequences of tasks on behalf of users. This development promises significant efficiency gains in everyday private life, for example in financial management or scheduling. At the same time, the use of such agents requires the disclosure of highly sensitive personal data, which creates a&#xa0;tension between maximising benefits and protecting privacy (privacy calculus). In this context, this article examines how different levels of perceived efficiency gains and data sensitivity affect the acceptance of AI agents. Based on an extended Technology Acceptance Model (TAM), a&#xa0;quantitative online study was conducted using an experimental vignette design. The results of the structural equation modelling (PLS-SEM) show that acceptance is highly context-dependent: while a&#xa0;high efficiency gain significantly increases the intention to use, perceived risks and data sensitivity act as strong inhibitors, which can, however, be mediated by trust. The article provides implications for the design of trustworthy human-AI interactions and discusses framework conditions for the use of autonomous agents.</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

Agentic AI zwischen Komfort und Kontrollverlust: Eine empirische Analyse der Akzeptanz autonomer KI-Agenten im Spannungsfeld von Effizienzgewinn und Datensensibilität

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年7月27日

查看全文
  • Elke Theobald,
  • Jule Bohe

摘要

The evolution of artificial intelligence (AI) is currently undergoing a paradigm shift from reactive chatbots to autonomous AI agents that are capable of independently planning and executing complex sequences of tasks on behalf of users. This development promises significant efficiency gains in everyday private life, for example in financial management or scheduling. At the same time, the use of such agents requires the disclosure of highly sensitive personal data, which creates a tension between maximising benefits and protecting privacy (privacy calculus). In this context, this article examines how different levels of perceived efficiency gains and data sensitivity affect the acceptance of AI agents. Based on an extended Technology Acceptance Model (TAM), a quantitative online study was conducted using an experimental vignette design. The results of the structural equation modelling (PLS-SEM) show that acceptance is highly context-dependent: while a high efficiency gain significantly increases the intention to use, perceived risks and data sensitivity act as strong inhibitors, which can, however, be mediated by trust. The article provides implications for the design of trustworthy human-AI interactions and discusses framework conditions for the use of autonomous agents.