With the growing integration of Artificial Intelligence (AI) in educational contexts, university students preparing for English language tests increasingly alternate between traditional search engines, such as Google, and large language models (LLMs) to support their test-related problem-solving. This study examines students perceptions of these tools, focusing on usability, efficiency, and their integration into English language test preparation workflows.Using a mixed-methods approach, we surveyed 140 university students from diverse academic disciplines and conducted in-depth interviews with 20 participants. Quantitative analyses, including ANOVA and chi-square tests, were employed to evaluate differences in perceived efficiency, satisfaction, and overall tool preference. The qualitative findings indicate that students frequently switch between GPT and Google depending on task demands, relying on Google for credible, multi-source information and rule verification, while using GPT for summarization, explanation, paraphrasing, and drafting responses for English test tasks. As neither tool alone was found to adequately support all aspects of English language test problem solving, participants expressed a strong preference for a hybrid solution. In response, we propose a prototype in the form of a chatbot embedded within a search interface, combining GPTs conversational strengths with Google reliability to improve English language test preparation and reduce cognitive load.

翻译：随着人工智能在教育场景中的日益融合，备考英语考试的大学生越来越多地在传统搜索引擎（如Google）与大型语言模型之间交替使用，以辅助其考试相关的问题解决。本研究考察了学生对这两种工具的认知，重点关注其可用性、效率及其融入英语考试准备流程的情况。采用混合研究方法，我们对来自不同学科的140名大学生进行了问卷调查，并对20名参与者进行了深度访谈。通过方差分析和卡方检验等定量分析方法，评估了学生在感知效率、满意度及整体工具偏好方面的差异。定性研究结果表明，学生会根据任务需求频繁切换GPT与Google：依赖Google获取可信的多源信息并进行规则验证，而使用GPT进行总结、解释、改写以及为英语考试任务草拟回答。由于发现任一工具均无法充分支持英语考试问题解决的所有方面，参与者表现出对混合解决方案的强烈偏好。为此，我们提出一个以聊天机器人形式嵌入搜索界面的原型设计，该设计结合了GPT的对话优势与Google的可靠性，旨在提升英语考试准备的效率并减轻认知负荷。