政府工作报告，为什么点名“高质量数据集”

Core Viewpoint - The article emphasizes the increasing importance of high-quality data in the development of artificial intelligence (AI), as highlighted in the 2026 government work report, which aims to foster a new intelligent economy and improve data resource utilization [3][4][5]. Group 1: Government Initiatives and AI Development - The 2026 government work report calls for the expansion of "AI+" initiatives, promoting the commercialization and large-scale application of AI in key industries, with the AI-related industry expected to grow to over 10 trillion yuan by the end of the 14th Five-Year Plan [4]. - The report reiterates the need to build high-quality data sets and improve the foundational data systems necessary for AI development [5][6]. Group 2: Data Quality and AI Training - High-quality data is essential for training AI models, with the article noting that the quality of data directly impacts model performance [6][7]. - As AI evolves from generative AI to physical AI, the demand for high-quality data becomes more critical, particularly for applications like smart driving and humanoid robots, where the complexity of required data increases [7][8]. Group 3: Challenges in Data Acquisition - The article discusses the challenges in acquiring high-quality data for physical AI, stating that while internet data is abundant, it is often unsuitable for training physical AI systems [9][10]. - The need for strong interactive data environments for embodied intelligence is highlighted, as traditional internet data does not facilitate necessary interactions [9][12]. Group 4: Potential of Private and Synthetic Data - There is significant untapped potential in private data across various industries, such as pharmaceuticals and fashion, which could provide high-quality insights for AI models [10][11]. - Synthetic data is identified as a promising area for development, with expectations for significant advancements in 2026, although the quality of synthetic data remains a concern [11][12]. Group 5: Data Standardization and Collaboration - The article points out the lack of a comprehensive data standardization system, which hampers data sharing and reuse across different manufacturers and sectors [13]. - There is a call for industry collaboration and innovation centers to address foundational data acquisition challenges and improve data quality [12][13].