How AI Can Solve The Rubik Cube Quickly

How AI Can Solve The Rubik Cube Quickly
Photo by Juan Pablo Serrano Arenas from Pexels

Have you ever placed solving a Rubik cube at least once in your life? Did it go well or bad? Did you solved the puzzle faster, or it takes you several minutes or even hours? Did you know this three-dimensional, multi-colored Rubik’s cube has been used and played in a competition since the 1970s?

There is no doubt that Rubik’s cube has caused a lot of people to furrow their brows for the past years. But considering today’s generation that is full of inventions, there is no surprise that hyper-intelligent machines can solve the Rubik’s cube in just a few seconds. 

The fastest time to solve a Rubik’s cube is three seconds. That was really amazing. But, what’s more, amazing is that the University of California at Irvine publicized that there is an AI system that can solve the puzzle in less than one second. 

The Artificial Intelligence system is called the DeepCubeA. It is a machine produced by the UCI professional mathematicians and computer scientists that features a reinforcement-learning algorithm. According to the said university, this device can solve the Rubik’s cube without knowing how the game works or without being manipulated by humans. 

What makes it more impressive is that it can make several moves to solve the Rubik’s cube. It can solve puzzles with six or even nine sides in less than seconds. Did you know anyone who can beat that record?

According to the professor of computer science, Pierre Baldi, in a statement published by the UCI, he stated that “It is possible for an artificial intelligence system to beat the world’s best Go and chess human players. But we noticed that Rubik’s cube and other more complex puzzles are difficult to be solved by a single computer. That’s when we thought that we can use AI approaches to solve the puzzle.” He also added that “they include more abstract, mathematical, and symbolic thinking to the system to ensure that the Rubik’s cube will be solved in no time. This means that there is a possibility that a machine that can solve the puzzle in just a second can become a system that can plan, reason, think, and do decision-making.”

The researchers of the said university released their conclusion in Nature Machine Intelligence. They also noted that the algorithm of the system they made was provided with more than ten billion puzzle combinations. The main goal of their system is to solve the complicated puzzles in less than 30 moves. 

The researchers also added that the DeepCubeA had determined the right strategy to solve the puzzle in the very shortest time possible. Aside from that, the system also passed 100 percent of test configurations. They also noted that the system could be used not only in solving the Rubik’s cube but also on other similar games, including Sokoban and Lights Out puzzles. 

Only the most intelligent person can only solve a Rubik’s cube using 50 moves. This is amazing, but the AI system will make you be more impressed since it can solve one puzzle using 20 moves only. According to the researchers, the system’s algorithm is dependent on the neural network. Neural network refers to the algorithms intended to determine the possible relationships of solving one puzzle by understanding how humans ingest and analyze the information given. 

Aside from that, the system is also reliant to machine learning techniques. This refers to a single system that enables the artificial intelligence approach to learn more by understanding the patterns and utilizing their analysis regarding the human interventions. 

But considering that the researchers only programmed the AI algorithm to solve complicated puzzles only, they do not know the exact reason how this had worked. To ensure that the machine will perform its work smoothly, the researchers isolated the DeepCubeA and trained it for two days. Within two days, the researchers had time to refine their skills by solving the Rubik’s cubes with several numbers of sides. According to one of the researchers, the machine acts as a human that learns by itself. He also added that the reasoning skills of the system are way different from the humans’. 

As released by the World Cube Association, during the early 1980s, the first Rubik’s Cube was solved by a human within 3.5 seconds. Though the UCI algorithm had impressed a lot of people, it is still not considered to be the number one conqueror to solve the cube quickly. The first robot that had solved the Rubik’s Cube in just 0.637 seconds was built in Germany and was named the Sub1 Reloaded. Last year, this record was beaten by the robot built by two American researchers and used a Rubik’s cube timer showing that the cube was solved for only 0.38 seconds. That was insanely incredible!

Final Thoughts

To sum it up, an AI algorithm can act as a human that can create their own strategies to solve a Rubik’s cube, provide suggestions that can improve the plan, mitigate the strategies that do not fit the plan look for possible alternatives that will work. With this, it gives the researchers with feedback that could help them determine the underlying process on how the DeepCubeA solves a Rubik’s cube within seconds. The researchers are looking forward to using the algorithm they made to teach many people the right way to solve a Rubik’s cube using fewer movements.

Rubik’s cube is a good puzzle game that is said to enhance the problem-solving skills of humans. But, if humans have learned the algorithms used by the AI system to solve the Rubik’s cube using lesser moves, do you still think it can help you improve your problem-solving skills and mental health?

There is no doubt that individuals are very intuitive that cannot be matched by any artificial intelligence machines. But it is no longer surprising that the innovated machines these days are way better in terms of their algorithmic rigor and computational power than humans.