It is universally acknowledged as a truth that an artificial intelligence-based system can perform calculations and other data-based analysis compared to humans. But when an AI creates something in hours, which took hundreds of years to find humans, then it is a new achievement.
Artificial Intelligence named Atom 2Vec, created by Stanford physicists has successfully differentiated between different atoms after observing the list of chemical compounds from a database.
AI used the ideas of the field of natural language processing in a particular, the idea is that the properties of words can be understood by looking at other words around them – cloning the elements according to their chemical properties.
JG Jackson at Stanford School and CJ Wood Professor Study Leader Shu-Cheng Zhang of Physics said, “We wanted to know that an AI can be smart enough to find periodic tables, and our team showed then do it “Of humanities and science.
Zhang says that the research published in the June 25 issue of National Academy of Sciences, the first important step towards one of his more ambitious goals, is designing the replacement of the Turing Test – the current gold standard for the gazing machine wisdom.
To pass an AI to the Turing Test, it should be capable of answering written questions in ways which are not separate from humans. Zhang thinks that the test is subjective. “Humans’ brains are surrounded by all kinds of irrationality. To pass the Turing test for an AI, it will need to reproduce all our human irrationalities.” “It is very difficult to do, not especially the good use of programmers.”
Zhang would like to offer a new benchmark of Machine Intelligence instead. They said, “We want to see that we can prepare an AI that can beat humans in search of new laws of nature.” “But to do this, we have to first check whether our AI can do some of the biggest searches already made by humans.”
By reviving the periodic table of elements, Atom 2vec has achieved this secondary goal, Zhang says.
Potassium is in the form of King. Zhang and his group modeled Atom 2 Wake on the AI program, which Google engineers built to analyze natural language. Called Word2Vec, Language AI works by converting words into numeric codes, or vectors. By analyzing the vectors, AI can guess the probability of the word appearing in a lesson after co-occurrence of other words.
For example, the word “king” often occurs with “man” by “queen” and “man”. Thus, the translation of the mathematical vector of “King” can be almost “king = a queen as a man with a woman.”
Zhang uses the same technique for atoms. “Instead of imparting any words and sentences from the collection of texts, we impart Atom 2vec to all known chemical compounds they are NaCl, KCl, H20, and so on.”
Elements, potassium (K) and sodium (Na) should have similar properties accordingly AI program because both the elements easily bind to chlorine (CL). Zhang said, “Like kings and queens are equal, potassium and sodium are the same.”
Scientists will use the idea of Atom 2vec for the discovery and designing new materials according to Zhang. Zhang said, “For this project, the AI program was unsafe, but you can direct it to give a goal and find it, for example, a material that is highly skilled in converting sunlight into energy is.”
His team is already working on another version of their AI program, which is 2.0, which will focus on breaking an unmistakable problem in medical research able to inspire immune response Molecules – which are specific to cancer cells. Currently, one of the most promising approaches to cancer treatment is cancer immunotherapy, which involves the use of antibodies that can attack the antigens on cancer cells.
Approximately 10 million different antibodies are produced by human body, in which every antibodies is combination of about 50 genes. Zhang says, “If we can map these building blocks to the mathematical vector, then we can arrange all the antibodies in something similar to the periodic table.” “Then, if an antibody is effective against an antigen and it is toxic, then there is a need of another antibody within the same family, which is effective but less toxic.”