文|《中国科学报》见习记者 赵婉婷 在体育界，迈克尔·乔丹是公认的伟大篮球球员；在机器学习领域，也有一个同名的闪耀存在，他被广 泛认为是学科先驱之一。前不久，他又多了一个新身份：中国科学院外籍院士。 迈克尔·欧文·乔丹（Michael I. Jordan）现任美国加州大学伯克利分校电子工程与计算机科学系、统计学 系杰出冠名教授，同时任北京大学、清华大学名誉教授。他曾获首届世界顶尖科学家协会奖，摘得约翰 ·冯·诺依曼奖等多项国际顶级荣誉，当选美国国家科学院、工程院、艺术与科学学院三院院士，并于 2025年11月当选中国科学院外籍院士。 在本科期间，乔丹学习的是心理学专业，对大脑如何工作的思考最终将他引入算法的探索道路。深耕机 器学习与统计推理领域40余年，乔丹还与产业界建立了广泛的合作。除了参与创办多家公司，他还曾任 支付宝、亚马逊的科学顾问。 "我对任何事物都感到好奇，尤其对不确定性感兴趣。"无论是基础研究还是产业应用，这位跨界教授都 秉持着为新问题出新点子的想法，从现实生活中寻找真问题，探索能够解决问题的新技术。 在近日举办的第十届世界华人数学家大会（ICCM）上，乔丹受邀作题为《人工智能的集体经济学视 角 ...

Core Insights - The article emphasizes the importance of focusing on the critical few rather than the trivial many to achieve extraordinary results, suggesting that success often stems from strategic adjustments rather than sheer effort [2][10][22] Group 1: Statistical Distributions - Normal distribution is commonly used to summarize data, but it often leads to misconceptions about the nature of success and outcomes [3][5][6] - Pareto distribution, or the 80/20 rule, illustrates that a small percentage of actions often leads to the majority of results, highlighting the need to concentrate resources on the most impactful areas [7][10][12] Group 2: Practical Applications - Personalized marketing efforts yield better results than mass marketing, as targeted communications resonate more with potential customers [11][12] - Investing in relationships with key individuals can provide greater returns than broad advertising strategies, emphasizing the value of quality over quantity in networking [13][18] Group 3: Customer Focus - Companies should prioritize their most valuable customers, as they contribute significantly to growth and success, rather than spreading resources thinly across all clients [14][15][16] - The concept of having a few deep, meaningful relationships is more beneficial than having numerous superficial connections [18][21] Group 4: Value of Unconventional Thinking - The article argues that true value lies in areas often overlooked by others, where small, intelligent actions can yield disproportionately large returns [22][23][25] - The perception of "luck" is framed as the result of being prepared and strategically positioned to seize opportunities when they arise [25][26]

Core Viewpoint - The book "Ladies' Tea: How Statistics Transforms Science and Life" by David Salsburg explores the evolution of statistics from its methodological origins to a systematic discipline, using the "Ladies' Tea" experiment as a narrative anchor [7][8]. Group 1: Historical Context of Statistics - The "Ladies' Tea" story originates from Ronald Fisher's 1935 work "Experimental Design," where a group of Cambridge scientists tested a lady's claim about the taste difference in tea preparation methods [8]. - The 19th century saw a dominant philosophical view of a "mechanical universe," where scientists believed that reality could be precisely described by a few mathematical formulas [8][9]. Group 2: Key Figures and Theories - Karl Pearson, in the late 19th century, proposed that experimental results should be viewed as distributions of numbers rather than precise measurements, leading to the development of statistical models [9]. - William Sealy Gosset, under the pseudonym "Student," focused on small sample sizes and introduced the widely used "t-test" [9][11]. - Fisher's work at the Agricultural Experiment Station led to the development of original data analysis tools, including "analysis of variance" and "randomized control" methods [11]. Group 3: Evolution of Statistical Methods - The 20th century saw the rise of hypothesis testing, with Fisher introducing the concept of the p-value as a measure of significance, although its interpretation remains debated [14]. - Non-parametric methods emerged as alternatives to traditional parametric methods, allowing for analysis without assuming a specific distribution [13]. Group 4: Philosophical Implications - The book raises philosophical questions about the application of statistical models in decision-making and the understanding of probability in real life [15]. - Despite the challenges and limitations of statistical methods, the discipline has become integral to various fields, including medicine and public policy [16][17].

Core Viewpoint - The article emphasizes the growing importance of statistics in the era of artificial intelligence, highlighting its applications across various fields such as business, medicine, engineering, and social sciences, while also addressing the challenges faced by AI technologies [1][2]. Group 1: Importance of Statistics - Statistics is not only used for government purposes but is also crucial in commercial, medical, engineering, and social science applications, focusing on data collection, analysis, and inference [1]. - The third National Conference on Statistics and Data Science featured over 600 academic reports, with nearly 25% related to machine learning and artificial intelligence [1]. Group 2: Challenges in AI - The "2024 Artificial Intelligence Development Report" by the China Academy of Information and Communications Technology identifies challenges such as insufficient interpretability of algorithms, security vulnerabilities, and irregular data labeling in AI [1]. - Current AI applications tend to prioritize algorithm functionality over understanding underlying mechanisms, particularly in high-stakes fields like medicine and construction, where stability and reliability are critical [1]. Group 3: Talent Development in Statistics - There is a growing emphasis on the cultivation of statistical talent in academia, as the demand for professionals in statistics and data science exceeds supply [2]. - The need for educational institutions to enhance their training capabilities is highlighted, aiming to produce more statisticians and data analysts while encouraging some to remain in academia [2]. Group 4: International Collaboration and Growth - Since the establishment of statistics as a primary discipline in 2011, there has been rapid development in statistical research in China, with Chinese authors now accounting for the second-largest share of publications in top international statistical journals [2]. - The conference attracted over 1,800 scholars, with 15% from abroad, indicating a growing international collaboration in the field [2].