AI is frequently overhyped, but one notable exception is its potential to unearth new materials.
These substances can provide the foundations for an array of transformative technologies, from life-saving drugs to new electronic devices. The discovery of graphene, for instance, could help solve the world water crisis, make smartphones virtually unbreakable, and even improve your sex life. However, the search for wonder materials can consume tremendous time and resources — with no guarantee of success.
AI could dramatically simplify the process.
Its primary advantage is speed. Computer models trained to identify patterns in vast chemical datasets could rapidly detect promising candidates.
A new AI tool developed by researchers at the University of Liverpool has demonstrated this capability.
The machine learning (ML) system was recently used to discover four new materials. These include a new family of solid-state materials that conduct lithium, which could be used to develop EV batteries.
In their study paper, the researchers describe their approach as “a collaborative ML-human expert workflow.”
First, a neural network model analyses the relationships between known substances to rank combinations of elements that could form new materials.
Scientists then use the rankings to guide their investigations in a targeted way.
In a statement, study lead author Matt Rosseinsky said the method blends the strengths of artificial and human intelligence:
This collaborative approach combines the ability of computers to look at the relationships between several hundred thousand known materials, a scale unattainable for humans, and the expert knowledge and critical thinking of human researchers that leads to creative advances.
Their tool joins a growing range of ML systems designed to unearth new substances. While wonder materials are traditionally discovered by humans in labs, the next graphene may be identified by AI.
Greetings Humanoids! Did you know we have a newsletter all about AI? You can subscribe to it right here.