TORONTO, July 15, 2019 /PRNewswire/ -- From May 20 to 23, the 2019 "Collision Tech Conference" was held in Toronto, Canada. As a cutting-edge technology conference with the fastest development in the North America, the event originated from Las Vegas and has been held six times this year. Nearly 26,000 participants, 3,750-odd CEO and 900-odd news media from 125 countries gathered in Toronto to discuss with science and technology leaders today's hot issues of science and technology.
Derek Haoyang Li, the founder and chief education technology scientist of Squirrel AI Learning by Yixue Group, an AI education company from China, was invited to the Collision Summit to share the scientific innovation achievements of their respective fields with representatives of Samsung, Micro-soft, Y Combinator and stars of show business.
Derek Li: Change education with AI
"Science and technology have changed every industry, but education has remained unchanged. I wish to change education thoroughly with AI," Derek Li said at Collision.
Squirrel AI Learning, established in Shanghai in 2014, is the first artificial intelligence company focusing on the education of primary and secondary schools in China, equipped with an AI adaptive learning engine based on senior algorithms and provides students with customized one-to-one education. Since its establishment, Squirrel AI Learning has raised nearly one billion yuan.
Squirrel AI Learning has established a Yixue AI Lab with SRI as the primary research partner and a Joint AI Lab with Carnegie Mellon University (CMU), appointed Professor Tom Mitchell, the god-father of machine learning and the dean of computer science of CMU as the chief AI scientist, and appointed Professor Ken Koedinger of the computer science and psychology system of CMU as chief learning scientist of Squirrel AI Learning. It has made a lot of investment in the research and development of AI education.
The achievements made in the past five years have made Derek Li confident enough to share his research achievements in the field of AI adaptive education before tens of thousands of participants at the conference.
He first introduced the three-level architecture of Squirrel AI Learning engine. The first level mainly comprises a learning map (a persona of the students) and a map of contents (including video teaching and answering questions). The second level comprises a target management engine, a user state platform and a recommendation engine, which can develop different learning strategies for different students. The third level is an interaction system which can interact with human users.
The architecture of Squirrel AI Learning system
Derek Li said he often got 95 points in his childhood. His hometown lies in inland China with a population of more than 100 million. The academic competition was fierce. In order to get one more point, he had to answer 10,000 more questions. "If there was Squirrel AI Learning at that time, I would have only needed to pay attention to the 5% of questions which I did not understand."
Compared with other world-leading adaptive education companies such as Dreambox, Knewton, a competitive advantage of Squirrel AI is its nano-level breakdown of knowledge points. For example, there are 300 knowledge pints in the math of junior high school. While other companies may break them down into 3,000 knowledge points, Squirrel AI can break them down into 30,000 ones.
Why is the breakdown so detailed? He gave an example, "If the pixels of a camera are increased by ten times, the photos will be clearer. Similarly, if the breakdown is detailed, we can target and diagnose the students' mastery of knowledge points more accurately and save time for them."
Other than imparting knowledge, Squirrel AI Learning has developed a system to enhance the students' learning ability. Derek Li called it MCM, which stands for model of thinking, capacity and method. The system can quickly detect and enhance the students' learning capacity and quality.