There is an old one The joke that physicists want to say: everything has already been discovered and reported in a Russian journal in the 1960s, we don’t know about it. Although hyperbolic, the joke accurately captures the current state. The amount of knowledge is growing huge and fast: the number of scientific articles posted on arXiv (the largest and most popular preprint server) in 2021 is expected to reach 190,000 – and this is just a subset of the scientific literature produced this year.
It is clear that we do not know what we actually know, because no one can read the entire literature even in their own narrow field (including journal articles, PhD thesis, lab notes, slides, white paper, technical notes, and reports). Indeed, it is entirely possible that in this mountain of paper, the answers to many questions are hidden, important discoveries are ignored or forgotten, and connections are hidden.
Artificial intelligence is a possible solution. Algorithms can already analyze the text without human supervision to find relationships between words that help uncover knowledge. But much more can be achieved if we move away from writing traditional theological scientific articles whose style and structure have rarely changed in the last hundred years.
Text mining has various limitations, including full access to paperwork and legal concerns. But most importantly, the AI does not actually understand their concepts and the relationship between them, and is sensitive to the bias of the data set, such as the selection of the documents it analyzes. For AI and, indeed, even for an experienced human reader – scientific research papers are difficult to understand in part because the use of jargon varies from one discipline to another and the same term can be used in different cases with completely different meanings. The growing interdisciplinary nature of research means that it is often difficult to define a topic accurately using a combination of keywords to discover all the relevant paperwork. Creating connections and (again) discovering similar ideas is difficult even for the bright mind.
As long as this condition exists, AI cannot be trusted and the output of AI after text-mining has to be double-checked, it is a tedious task that denies the purpose of using AI. To solve this problem, our science papers need to be machine-readable, not machine-readable.Understandable, Writing (rewriting) a special type of programming language. In other words: teach science to machines in the language they understand.
Writing scientific knowledge in a programming-like language will be dry, but it will be sustainable, because new ideas will be added directly to the library of science that machines can understand. Plus, since machines are taught more scientific information, they will help scientists simplify their logical arguments; Spot errors, inconsistencies, theft and forgery; And highlight connections. Regarding physical law, AI is more powerful than AI trained only on data, so science-intelligent instruments will be able to help future discoveries. Machines with a great knowledge of science can help instead of human scientists.
Mathematicians have already begun this process of translation. They are teaching mathematics on computers by writing theorems and proofs in languages like Lien. Lin is a proof assistant and programming language in which one can introduce mathematical concepts in the form of objects. Using familiar objects, Lynn could argue that a statement is true or false, thus helping mathematicians verify the evidence and identify their arguments insufficiently rigorous. The more Lynn knows math, the more she can do. The Xena project at Imperial College London aims to input a complete undergraduate math curriculum. One day, proof assistants can help mathematicians verify their reasoning and research by exploring vast knowledge of mathematics.