Squirrel AI Learning Provides Answers for China’s Education Dilemmas at “the Fire of Innovation” Hosted by Geekpark
The Fire of Innovation“ was held by GeekPark was held at the Computer History Museum in Mountain View, Silicon Valley. Here, GeekPark and Liepin jointly released their list, “Geek Employers @ China 2018“. The list included 50 technology-driven companies, covering many emerging forces that optimize each industry and change the world with innovative technologies in various fields.
A key question raised at the event was “How to implement “AI + education” to truly promote the progress of education?“Yixue Squirrel AI Learning gave their answer at The Fire of Innovation: technology is the best way to address the core pain points faced by education today.
As a representative of AI-powered education among the companies in “Geek Employers @ China 2018″, Squirrel AI Learning successfully developed China’s first AI adaptive learning engine based on advanced algorithms with complete proprietary intellectual property rights. Squirrel AI Learning is also China’s first AI company to apply AI adaptive learning technology to K12 education. At the event, Joleen, a partner at Squirrel AI Learning, shared her latest thinking about AI education with the attendees.
The full content of the speech is provided below:
In education, the value of a technology or methodology is judged not only by the resulting improvements in efficiency and effectiveness, but also by its universal applicability, i.e. the extent to which this technology can be used in education, a fundamental sector of the society’s infrastructure. Only if it can have an effect on more students from different backgrounds with different bases can it really promote the progress of education.
There are thousands of cities and regions in China today. Economic development varies greatly in first-tier, second-tier, third-tier, fourth-tier and fifth-tier cities. The distribution of education resources has become more and more imbalanced, particularly in K12 education. Education resources are imbalanced. All education resources are concentrated in key primary schools, middle schools and high schools. Difficult admission to good schools, long commutes and other problems concern every parent. With the rise of the Internet, online education can alleviate the imbalance in education resources to some extent. However, due to the differences in regional development, students have different learning habits and perceptions of the Internet, which is a new pain point for online education.
Simply speaking, the expensiveness of houses in school districts implies the seriousness of the imbalance in education resources in China.
AI is the technology that has brought the biggest impacts and changes to traditional industries. The search for a right way to integrate AI into education started early. There are excellent cases of seeking solutions to the fundamental problem of uneven distribution of education resources. Squirrel AI Learning is a typical case.
Founded in 2014, Yixue Squirrel AI Learning is the first AI company in China to apply AI adaptive learning technology to K12 education. Squirrel AI Learning successfully developed China’s first AI adaptive learning engine based on advanced algorithms with complete proprietary intellectual property rights. It has opened more than 1,000 AI-powered learning centers in more than 200 cities across China, with more than 1 million students.
What AI Can Change
In recent years, AI has produced profound impacts on different industries, both worldwide and within China.
In such fields as intelligent assistant, face recognition, automated driving, content distribution and so on, AI has brought changes and optimizations that previous technologies could not achieve. AI has achieved omniscience and omnipotence, recognition of the whole from one small clue, infinite computing power and self-evolution. Among all, Squirrel AI Learning that targets China’s education sector is a good story.
First, AI is versatile and knows-it-all. In just 17 seconds, IBM Watson can read tens of thousands of medical documents and patient cases, which will take a human doctor a lifetime to read. If an automated driving system knows all streets and roads, all places of departure and destination for every car and all road maps, there will be no need to worry about accidents involving young drivers and tired drivers. Even traffic congestion and accident rates will be largely reduced. Squirrel AI teaching robot has known hundreds of thousands of knowledge points from primary school to high school, and has learnt the correlations between millions of exercises and those knowledge points. With omniscience and omnipotence, our AI system scans each student’s user portrait and knowledge points, to know exactly what a student knows and what he doesn’t.
Second, AI has the feature that small things can tell the big stories. When there was no AI a few years ago, user portraits were pretty crude. With AI, user portraits can be 10 times or even 100 times more precise. For example, Toutiao not only can accurately push a user’s favorite news about technology, education, health, entertainment, but also can recommend content more precisely if a user only wants to read news and articles about B-to-C companies that raise more than 100 million RMB funds. The same is true for AI in the education sector. After a comprehensive scan by our AI system, students don’t need to do so many exercises. The learning speed and mastery of each knowledge point vary from student to student. AI can distinguish in real time and analyze in detail to match the knowledge points each student needs to know in a personalized manner.
Third, AI has infinite computing power. Using sophisticated AI models, hedge funds have achieved a 20-year compound annualized yield of up to 35%, surpassing Mr. Buffet. The infinite computing power of Squirrel AI is reflected in the nano-scale breaking down of knowledge points. Our system has broken 500 knowledge points in junior high school to 30,000 knowledge points. Students’ learning efficiency can increase 10 times and the time wasted can reduce for 10 times.
Fourth, AI can be self-evolving. In Japan, there are already unmanned hotels, where robots have replaced humans in all service positions. Microsoft Xiaoice has passed the Turing test by singing. The poems created by it are very humanized and even perfect. In self-evolution, Squirrel AI‘s teaching system originated from senior and elite teachers. But now, its teaching level has surpassed elite teachers and senior teachers. In teaching, AI already can do a lot of human teachers’ work. Human teachers can focus more on curriculum development and moral education. Squirrel AI‘s teaching robot has done better than human teachers in performance improvement in all four human vs. machine competitions since 2017. In the fourth human vs. machine competition in 100 cities, compared with the human group, the AI group’s score increased 5 points; the AI group learnt 42 knowledge points on average, while the human group learnt 28 knowledge points.
China’s online education has gone through six tides.
The first tide is the Internet+ education. Internet+ brings offline classes online. It can popularize the best education resources regardless of time and place. However, the completion rate is low and the dropout rate is high. Many Internet+education platforms have transformed into the fifth and sixth tides as well. The second tide is the tool-based education model. Due to the rigid demand for searching exercises and finishing homework, it attracts lots of free traffic and users. However, tool platforms can’t get involved in the real teaching process. The third tide is O2O. O2O connects idle teaching resources. However, the quality of teachers is uneven, so there is no guarantee that every student can receive high-quality education. The fourth tide is the content-based education model. High-quality courseware content can be very effective in attracting users, but big platforms can invest in curriculum development on their own or work with content providers to catch up with and surpass content-based platforms. The fifth tide is one-on-one tutoring by real teachers. Real-time one-to-one online teaching can solve the problem of poor focus and poor continuity in online learning. However, when the scale expands 100 times, the cost of teachers also goes up 100 times.
In the end, the sixth tide has come, which is AI+ education. Squirrel AI intelligent adaptive learning is a typical case of AI+ education. Squirrel AI intelligent adaptive teaching engine uses many algorithms, such as the genetic algorithm, neural network technology, machine learning, graph theory, probabilistic graph model, logistic regression model, knowledge space theory, information theory, Bayes theory, knowledge tracking theory, education data mining, learning analysis technology and so on. Through these algorithms, the system can plan the best learning path, maximize the learning efficiency, recommend the most suitable learning content based on the individual preferences, learning habits and styles of different students, and dynamically evaluate students’ ability level in real time. Finally, it can accurately locate students’ current knowledge state and predict their future scores.
Squirrel AI Learning is using AI technology to optimize education efficiency and education resource allocation efficiency to maximize the optimization of and progress in education.
In the meantime, Squirrel AI Learning has established a Joint Lab for AI in Silicon Valley with Stanford Research International (SRI) and a Joint Lab for AI Intelligent Adaptive Education with the Institute of Automation, Chinese Academy of Science. Since 2017, it has published several papers on AI intelligent adaptive education, trying to solve the fundamental problems in K12 education in China from both industrial and academic fields.
Squirrel AI Learning is a 3-in-1 company that integrates teaching content, an intelligent adaptive education platform and online + offline service, where most European and American companies only have one or two out of the three components.
Currently, the challenge for Squirrel AI Learning is to recruit more top AI talents and education experts to join the teamand continue to optimize and improve the algorithms and models behind the Squirrel AI intelligent adaptive teaching system.