港科大教授实测AI眼镜“作弊”：30分钟碾压95%的学生，把传统教学评估体系整破防了

Core Viewpoint - The article discusses a recent experiment at Hong Kong University of Science and Technology where an AI-powered glasses equipped with ChatGPT-5.2 took a final exam, achieving a score of 92.5, outperforming over 95% of human students, raising questions about the validity of traditional educational assessment methods [1][4][6]. Group 1: Experiment Overview - The AI glasses, developed by a team led by Professors Zhang Jun and Meng Zili, were designed to "cheat" in a controlled exam environment for the course "Computer Network Principles" [7][8]. - The AI glasses utilized a combination of hardware and software, including a camera for capturing exam questions and a cloud-based model for generating answers [13][12]. - The experiment aimed to evaluate the performance of AI in a traditional academic setting, highlighting the potential challenges to existing educational evaluation systems [5][6]. Group 2: Performance Metrics - The AI glasses scored 92.5, with perfect scores in multiple-choice and single-page short answer questions, and a high score in multi-page short answer questions [14]. - The performance demonstrated strong reasoning capabilities, even in complex questions that required contextual understanding [14][15]. - The experiment revealed that the AI could effectively complete the reading, understanding, and answering process, raising concerns about the relevance of traditional assessment methods [22][23]. Group 3: Implications for Educational Assessment - The success of the AI in the exam challenges the effectiveness of current educational assessments that focus primarily on standardized answers [22][33]. - The article suggests that traditional assessments may not adequately measure critical skills such as problem-solving, creativity, and understanding, which are essential in real-world scenarios [39][43]. - There is a growing need to shift the focus of educational evaluations from merely providing correct answers to assessing the reasoning processes and understanding behind those answers [40][48].