Adaptive Learning Engine Acts as an AI Super Teacher for K-12 Students

May 27, 2019 – As I visited the many exhibits on display at Collision 2019 last week in Toronto, I was approached by a young man who saw my press pass and invited me to meet with Drek Haoyang Li, the Founder and Board Chairman of Squirrel AI Learning. Headquartered in Shanghai, Squirrel AI Learning has been developed by Yixue Group, a company whose expertise lies in adaptive learning systems. The choice of the squirrel as the company’s symbol relates to how the animal is seen in China as a symbol of agility, diligence, and management.

Squirrel AI Learning has at present put in place adaptive learning after and in-school programs in subjects such as mathematics, physics, chemistry, and English and Mandarin language learning. It is being used in 1,700 schools, and more than 200 cities across China and has presence in North America in an artificial intelligence education laboratory in New York City as well as in a jointly managed learning lab in Silicon Valley where the company is collaborating with Stanford, Carnegie Mellon, and UC Berkeley universities. In China, at present more than 3,000 teachers are using Squirrel Ai Learning to help their students.

The Chief AI Officer of the company is Tom Mitchell, Dean of the School of Computer Science at Carnegie Mellon University, and a pioneer in AI and machine learning. Mitchell’s research has focused on statistical learning algorithms for discovering how our brains represent information. He is the creator of NELL, the Never-Ending Language Learner, a software program that teaches itself to read by scouring unstructured web content.

Mitchell believes the algorithms behind Squirrel AI demonstrate a high degree of adaptability to the way individual students learn, that the technology shows that it is useful to identify current levels of student understanding through interaction with them, and over time adapt and develop best ways to maximize learning potential.

This is what makes Squirrel AI unique, a capacity to learn about the learner and then through interaction design optimal methods of learning to match what works best with an individual. Squirrel AI doesn’t stuff a student’s head with facts. Instead it takes an incremental strategy to build learning success. If a student is achieving 70% in a particular subject, Squirrel AI Learning will then map out a strategy to increase that grade to 75%.

When tested against human teachers in China recently, Squirrel AI demonstrated on four different occasions better outcomes than human-taught groups by a margin of nearly 5%. In December of last year, the company reported its adaptive learning technology was being used in China to teach over 1 million children of poor families for free. As the company develops its Open Squirrel AI platform, the next evolution of the engine, it hopes to gain wider usage globally and foster a new generation of self-motivated learners. It certainly is off to a good start.