An artificial intelligence that can predict how a piece of glass responds to heat and pressure could one day also be used to model traffic flow.
While most solid materials have a regular atomic structure, the atoms in glass have a more irregular arrangement, resembling a liquid that has been frozen in place. Physicists have long wanted to know more about this 鈥�glass transition鈥�.
鈥淕iven that glasses are everywhere 鈥� from windows to your phone screen 鈥� it鈥檚 odd that we don鈥檛 really understand its structure and dynamism,鈥� says Victor Bapst at AI firm DeepMind.
Advertisement
Bapst and his colleagues used machine learning to model all the ways that the particles in a piece of glass respond to different temperatures and pressures. The AI was trained to predict the future of one particle and that of its immediate neighbour particles.
By running the software several times to take into account all the various combinations of particles and neighbour particles, the AI was able to model how the entire piece of glass would react to different conditions.
The AI鈥檚 predictions of initial particle movements under different pressures and temperatures were 96 per cent accurate, on average. But this decreased to 64 per cent over longer time scales 鈥� when the glass particles are moving over longer distances, as if they were in a liquid, this makes it harder to predict what happens. In both cases, the AI was more accurate than current computer simulation methods.
The researchers hope to use this AI to model traffic flow, treating cars as particles and using the same concept of neighbour particles to predict what happens to cars stuck in a traffic jam.
Nature Physics
Want to get a newsletter on all the latest advances in AI? Register your interest and you鈥檒l be one of the first to receive it when it launches.
