Team of researchers at the University of California have designed an artificial intelligence system that can solve a Rubik’s Cube in just over a second.
Invented in 1974, only few people have been able to solve the 3D logic puzzle. By means of the algorithm named DeepCubeA, the process of solving becomes no longer taxing.
“It learned on its own,” commented the author of the report, Prof Pierre Baldi. Furthermore, he also added that the AI applied a completely different logic in comparison to that used by humans to solve the puzzle.
Baldi is a professor of computer science at the University of California, Irvine.
According to reports, the algorithm was fed with 10 billion different combinations of the puzzle and has set a target for the algorithm to be able to decode all the combinations within 30 moves.
The algorithm was then tested on 1000 puzzles and was able to solve all of them. Moreover, it chose the shortest path, almost 60% of the time.
In comparison, the fastest, a human could solve the puzzle was in maximum 50 moves. The AI system however solved it in an average of 28 moves.
However, the latest algorithm is not the first one to be able to solve the puzzle the fastest. A previous algorithm model of the Massachusetts Institute of Technology (MIT) could solve it three times faster, reports suggest.
However, the MIT model did not use the neural network, in other words, it did not rely on how the human brain works, neither did it mimic the machine learning technique and it was only programmed to solve the puzzle. In another incidence in 2018, a robot could solve it in mere 0.38 seconds.
The latest system stands out in the fact that it can teach itself to tackle the challenge. The newest system also concentrates efforts towards building AI that is beyond games and can be made capable to solve real world problems.
“The solution to the Rubik’s Cube involves 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,” commented Prof Baldi. “How do we create advanced AI that is smarter, more robust and capable of reasoning, understanding and planning? This work is a step toward this hefty goal.”
The study was summarized in Nature Machine Intelligence.