人工智能专题:DeepSeek时代:让AI更懂中国文化的美与善

Core Insights - The report focuses on the DeepSeek model and its application in understanding Chinese culture, particularly in the arts, highlighting its advantages in Chinese language scenarios and its performance in various assessments [1][2]. Group 1: DeepSeek Model Performance - DeepSeek has shown outstanding performance in Chinese language contexts, ranking highly on the SuperCLUE open-source leaderboard [1][9]. - In creative English story writing evaluations, DeepSeek is positioned in the top tier [1]. Group 2: AI Applications in Art - AI is widely applied in the art sector, with examples including the MIT-IBM Watson AI Lab and Harvard Art Museum, which have collected over 53 million machine-generated descriptions and labels covering approximately 378,000 art images [1][10]. - Microsoft and the Dutch Bank have utilized GAN technology to restore artworks, generating new pieces in the style of Rembrandt [1][11]. Group 3: Zhejiang University Initiatives - Zhejiang University has undertaken numerous projects combining culture and AI, including the "Comprehensive Collection of Chinese Paintings," which is a significant national cultural project supported by various government departments [1][36]. - The project encompasses over 12,250 pieces of artwork, making it the most comprehensive collection of its kind, with contributions from both domestic and international institutions [1][37]. Group 4: Educational and Research Support - The university's School of Art and Archaeology leverages its multidisciplinary strengths to undertake national-level projects and offers general education courses to enhance students' understanding of classical painting and culture [1][36]. - A national-level database for traditional Chinese paintings has been established, featuring an analysis system with capabilities for multi-classification, custom retrieval, color analysis, and creative generation [1][36]. Group 5: Ethical and Aesthetic Considerations - The course discusses the normative use of large models in the Software 3.0 era, emphasizing the importance of aesthetics, ethics, education, culture, and values in the development of AI models [1].