Bite-sized Breakthroughs: AI Solves a Rubik’s Cube
Eren Bilen and Pranav Azad ’27. Photo by Dan Loh.
Student-faculty data analytics project looks at teaching AI to solve the Rubik’s Cube like a human
by Tony Moore
Curiosity drives everything at Dickinson. In labs, studios, archives and communities around the world, students and faculty are asking questions big and small—and often finding surprising answers. Here's a look at a recent query.
Department: Data Analytics
Researchers: Pranav Azad ’27 (computer science, data analytics) and Assistant Professor of Data Analytics Eren Bilen
Project:
“AI for Rubik’s Cube: Mimicking Human Solve”: Creating a human-like AI Rubik’s Cube solver. Azad trained the AI on thousands of human solves and developed it into an opponent users can race against.
Why it matters:
As Bilen explains: “Most Rubik’s Cube solvers are designed to find the most efficient solution, but that is not how people actually solve. Human cubers rely on methods they have practiced, recognize familiar patterns and sometimes make less-than-perfect choices. We trained an AI model we call Kybos on thousands of human solves so it could learn to solve in the same way. This created an AI opponent that feels more like another person than a machine.”
Players can race against Kybos anytime at kyboscube.com.
Related video: Radiath Kamal Patwary '28 Races the Clock, One Cube at a Time
TAKE THE NEXT STEPS
Published September 8, 2026