China is developing an AI-based intelligence platform called “Supermind” to track millions of scientists and researchers worldwide, aiming to hoover up breakthrough technologies for industry and the military. The state-funded platform uses sophisticated AI systems to help find talent for China and is under construction in a new “information and intelligence” center in Shenzhen, home to big tech firms like Huawei, ZTE and Tencent, Newsweek reported. The project, fueled by a US$280 million investment mainly by the Shenzhen government, represents a significant AI-empowered step in China’s bid to win a global technology competition with the US. The Supermind will offer users 300 million global science and technology research papers and 120 million patents, and have the ability to locate 130 million international scholars or “human talents” to scrutinize their work. The system will be constantly updated and include Hong Kong and Macau in its networks. https://asiatimes.com/2024/03/us-china-in-a-tech-tracking-supermind-competition/ The article is fairly long and goes on to describe a similar project by the US' Defense Innovation Unit. AI applied to science information is certainly a needed innovation. The conventional model of scientific information is that scientists do their work and then publish results as refereed articles in the most prestigious journal they can get them accepted. The results may or may not be right, and they are rarely replicated. Determining whether the results "fit" with the existing body of knowledge is left to the referees, typically volunteers or poorly paid junior scientists. The consolidation of results into an organized and coherent body of knowledge is done informally by academic scientists who compile class notes and write books. This is all rather clunky and inefficient, particularly as there has been tremendous growth in scientific publication. AI offers the possibility of creating a coherent body of knowledge from the existing literature and then incrementally adding in new publications with much better tests for coherency, probably accuracy, importance, and reliability. Besides publications, the scientific/technical literature also includes lots of other information, such as methods for doing various experiments, catalogs of manufactures of scientific instruments, components, devices, reagents, etc. Indexes of who is doing what in various fields and institutions. This could lead to a major advance in how science and technology advance. As well as how scientific/technical espionage is done, as described in the article.
