New AI Program Reproduces the Periodic Table in Just a Few Hours

Stanford University scientists use AI to arrange the periodic table.

A team of physicists at Stanford University recently developed an artificial intelligence (AI) program that arranged the 118 elements in the periodic table according to their properties in only a few hours.

A feat of chemistry and ingenuity, the periodic table of elements took almost a century to organize, with various human scientists making several attempts at the task over time. An AI program called Atom2Vec arranged the table in just a few hours.

Developed by Stanford physicists, the AI was able to distinguish between different atoms by analyzing chemical compound names from an online database. It used natural language processing concepts to group the elements together according to their chemical properties. This means that the properties of words can be understood by assessing the other words associated with them.

The Atom2Vec was modeled from a Google program called Word2Vec, which works by converting words into numerical codes, or vectors. By analyzing the vectors, the AI can estimate the probability that a word will appear in a text given the co-occurrence of other words.

By using these concepts, machines now have a greater chance of successfully analyzing and understanding human language. “We wanted to know whether an AI can be smart enough to discover the periodic table on its own, and our team showed that it can,” said study leader and Stanford physics professor Shoucheng Zhang.

According to Zhang, the Atom2Vec is the first step toward a bigger goal—replacing the Turing test.

The Turing test was devised back in 1951 and is still the current standard for gauging machine intelligence, which is essentially determining whether it can respond to questions that are indistinguishable from those of a human. Zhang believes the test is flawed because it has subjective elements.

“Humans are the product of evolution, and our minds are cluttered with all sorts of irrationalities. For an AI to pass the Turing test, it would need to reproduce all of our human irrationalities,” Zhang said. “That’s very difficult to do, and not a particularly good use of programmers’ time.”

The team believes that the Atom2Vec’s success can introduce a new benchmark of machine intelligence. “We want to see if we can design an AI that can beat humans in discovering a new law of nature,” Zhang continued. “But in order to do that, we first have to test whether our AI can make some of the greatest discoveries already made by humans.”

Zhang hopes that future scientists can use the knowledge from Atom2Vec to discover and design new materials. “For this project, the AI program was unsupervised, but you could imagine giving it a goal and directing it to find, for example, a material that is highly efficient at converting sunlight to energy,” Zhang said.

The team is already working on a new version of the AI program that will focus on designing antibodies to attack antigens in cancer cells.

The research for Atom2Vec was published in the June 25 issue of Proceedings of the National Academy of Sciences.

For more breakthroughs in AI, check out how this Industrial Robot Meets Artificial Intelligence to Create Art.