近日，随着高考落下帷幕，上海人工智能实验室旗下司南评测体系OpenCompass，所公布的AI高考全卷评测结果引起了社会各界的广泛关注。这次评测是一次颠覆性的探索，它首次全面检验了大模型在“语数外”全卷方面的能力。
据悉，此次评测共选取了7个大模型进行能力测试。在满分为420分的高考全卷中，阿里通义千问2-72B脱颖而出，以最高分303分成为此次评测的冠军。OpenAI的GPT-4o紧随其后，得分296分。上海人工智能实验室的书生·浦语2.0排名第三，三个大模型的得分率均超过70%，显示出人工智能在高考中的显著实力。
值得一提的是，评测结果显示，在数学科目上，所有参与评测的大模型均未达到及格水平。这也反映出人工智能在处理复杂数学问题时的挑战与不足。尽管如此，人工智能在其他科目上的表现仍然引人注目，尤其是自然语言处理方面展现出了强大的能力。
此次AI高考全卷评测结果的发布，不仅揭示了人工智能在高考中的表现，也为未来人工智能技术的发展提供了重要的参考依据。同时，这也引发了关于人工智能在教育领域应用的广泛讨论和期待。未来，随着人工智能技术的不断进步，其在教育领域的应用也将更加广泛和深入。
（新闻作者：XXX）
英语如下：
News Title: “AI First National College Entrance Examination Test Results Revealed: Top Models Achieve High Scores, but Face Challenges in Math”
Keywords: AI College Entrance Examination Evaluation, Big Model Competition, Scores Published
News Content:
The results of the first AI National College Entrance Examination full-length assessment have been announced by the OpenCompass evaluation system of Shanghai AI Labs, sparking widespread attention from all sectors of society. This assessment is a groundbreaking exploration that comprehensively tests the abilities of large models in all subjects including language, mathematics, and foreign languages.
It is reported that seven large models were selected for this assessment. In the full-length exam with a maximum score of 420 points, Alibaba’s Tongyi Qianwen 2-72B stood out with a highest score of 303 points, becoming the champion of this assessment. OpenAI’s GPT-4o followed closely with a score of 296 points. The third place was taken by Shanghai AI Lab’s Shusheng PuYu 2.0, with all three major models achieving a score rate exceeding 70%, demonstrating significant AI capabilities in the college entrance examination.
It is worth mentioning that the assessment results showed that none of the participating AI models achieved a passing grade in mathematics. This reflects the challenges and shortcomings of AI in dealing with complex mathematical problems. Nevertheless, AI’s performance in other subjects was still noteworthy, particularly in natural language processing where it demonstrated strong capabilities.
The publication of the results of the AI National College Entrance Examination assessment not only reveals AI’s performance in the exam but also provides an important reference for the future development of AI technology. It has also sparked widespread discussion and expectations about the application of AI in the field of education. As AI technology continues to progress, its application in education will become more extensive and deeper.
(Author: XXX)
【来源】https://www.yicai.com/news/102156889.html
Views: 1