The Rubik's Cube, a colored 3D puzzle block that challenges a person's problem solving skills, has always been the subject of several experiments in the past.
In a new study, an artificial intelligence system
developed by a group of researchers can solve the Rubik's Cube, however it's jumbled, faster than any human could.
Self-Taught AI Solves The Rubik's Cube
Computer scientists and mathematicians at the University of California, Irvine, programed an AI that can solve the Rubik's Cube superfast — a little over a second. Called the DeepCubeA, this AI is nothing new. In fact, there are other systems in the past that solved the Rubik's Cube faster than the DeepCubeA.
What's so special about it? For starters, the DeepCubeA isn't limited to just solving the Rubik's Cube. Previous systems that successfully finished the puzzle were solely made for this purpose alone, meaning their functions are limited, unlike the DeepCubeA, which can be possibly used in other domains in the future.
In addition, these systems were able to solve the 3D puzzle with the help of human-scripted algorithm.
DeepCubeA, on the other hand, does not require human intervention. Through an approach called reinforcement learning, this AI taught itself how to solve the Rubik's Cube using minimal moves in the fastest time possible.
Reinforcement learning encourages the AI to achieve a targeted goal through some sort of reward system. It earns points for each successful strategy. It'll lose points for ineffective moves.
Aside from reinforcement learning, DeepCubeA also relies on neural network
— similar to how a human brain works.
"The solution to the Rubik's Cube involves more symbolic, mathematical and abstract thinking, so a deep learning machine that can crack such a puzzle is getting closer to becoming a system that can think, reason, plan and make decisions," Pierre Baldi, senior author and UCI Distinguished Professor of computer science, said
in a press release.
The study is published
in the journal Nature Machine Intelligence