SoylentNews is people
SoylentNews
Stanford AI recreates chemistry's periodic table of elements
It took nearly a century of trial and error for human scientists to organize the periodic table of elements, arguably one of the greatest scientific achievements in chemistry, into its current form. A new artificial intelligence (AI) program developed by Stanford physicists accomplished the same feat in just a few hours.
Called Atom2Vec, the program successfully learned to distinguish between different atoms after analyzing a list of chemical compound names from an online database. The unsupervised AI then used concepts borrowed from the field of natural language processing – in particular, the idea that the properties of words can be understood by looking at other words surrounding them – to cluster the elements according to their chemical properties.
[...] Zhang and his group modeled Atom2Vec on an AI program that Google engineers created to parse natural language. Called Word2Vec, the language AI works by converting words into numerical codes, or vectors. By analyzing the vectors, the AI can estimate the probability of a word appearing in a text given the co-occurrence of other words.
[...] Zhang hopes that in the future, scientists can harness Atom2Vec's knowledge 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.
Wake me up when an AI discovers the Island of Stability.
Learning atoms for materials discovery (open, DOI: 10.1073/pnas.1801181115) (DX)
(Score: 3, Insightful) by Anonymous Coward on Thursday June 28 2018, @09:23AM (6 children)
If my memory does not fail me, I thought that when the initial periodic table was assembled, some spots were
empty. Those elements weren't identified at that time yet. The AI had all this information already available to use.
(Score: 4, Informative) by Anonymous Coward on Thursday June 28 2018, @10:36AM (4 children)
I was excited when I saw this on phys.org, so I read most of their blurb there.
It has nothing to do with the way that humans got the periodic table.
Humans had a bunch of substances with arbitrary names that they then combined in various ways, figured out what interacts with what, and then concluded that some of them were fundamental and some of them composite, and then figured out that the fundamental ones could be classed into groups based on how they interacted with the other ones.
This code was given the chemical notations for various substances, and was taught chemical reactions, and it then figured out that basic substances exist and they can be classed into groups. Well, duh. it wouldn't have taken us a thousand years to do it if stuff was labeled Ca2CO3 and H2O and CH4.
(Score: 3, Insightful) by OrugTor on Thursday June 28 2018, @02:30PM (2 children)
My thought too. I was hoping It would be given the Hamiltonians, solve and discover stable configuration of electrons and deduce elements and their chemical properties. Instead we get another chatbot on steroids.
BTW what is Ca2C03?
(Score: 2) by Taibhsear on Thursday June 28 2018, @02:36PM
A typo for CaCO3, I'm guessing.
(Score: 0) by Anonymous Coward on Thursday June 28 2018, @03:38PM
I'm sorry, it's been twenty years since I needed to know these things... I was thinking of CaCO3
https://en.wikipedia.org/wiki/Calcium_carbonate [wikipedia.org]
(Score: 3, Insightful) by frojack on Thursday June 28 2018, @05:02PM
Its easy when you know what your starting point is, and where your desired end point is.
People knew none of this! People didn't even know what they were looking for exactly until well over half the elements were already cataloged in tabular form.
Best possible use of this is figuring out some alien life form's words for various elements.
Most likely use: someone's dissertation.
No, you are mistaken. I've always had this sig.
(Score: 0) by Anonymous Coward on Thursday June 28 2018, @11:39PM
Yeah. It would be interesting what it could predict if it were trained using only the information available in 1780 (long before the discovery of nuclear physics).