随着人工智能在信息检索过程中的应用日渐深入，区别于基于互联网技术的搜索引擎，基于深度学习和自然语言处理的大语言模型技术不断发展成为一种的搜索新范式。本研究引入认知需求水平这一变量，探究了在搜索引擎和大语言模型两种范式下，认知需求水平对个体信息搜索行为的影响。研究结果显示，搜索技术范式对信息搜索行为以及信息搜索的质量有显著影响，大语言模型范式下的个体在积极预期违背水平、广告察觉、愉悦感知和信息无用性感知方面的表现更为积极，在信息搜索质量上也存在优势，尤其是在输入文本长度和完整度方面。虽然认知需求水平仅在信任感知方面表现出边缘显著的主效应，但搜索技术范式和认知需求水平对于信息无用性感知存在显著的交互作用。本研究通过分析行为数据比较不同认知需求水平的个体如何通过不同的搜索范式进行所需信息的获取，为人机交互中个体信息搜索的需求与偏好提供了实证依据，并希望据此为信息搜索中人工智能的更广泛、更深度应用提供实证建议。

The application of artificial intelligence in the process of information retrieval is becoming increasingly profound, with the development of large language models representing a new paradigm shift distinct from internet-based search engines. This study introduces the need for cognition (NFC) and comparatively explores the impact of NFC level on individual information search behavior under two paradigms: search engines and large language models. The results indicate that the search technology paradigm has a significant influence on information search behavior and quality. Individuals under the large language model paradigm exhibit more positive attitudes and behaviors across multiple dimensions. And there is a significant interaction between the search technology paradigm and NFC level in the perception of information uselessness. This study compares how individuals with different NFC levels acquire needed information through different search paradigms, providing empirical evidence for individual information search needs and preferences in human-computer interaction. The study also aims to offer empirical suggestions for the broader and deeper application of artificial intelligence in information search.