AIED2019: How the World’s Top AI Experts Apply AI to Education

AI has become the core driving force of the new round of industrial transformation and ushered in a new period of industrial explosion. How to break through the barriers between technology and industry has become a heated issue. AI in education has become a favorite target for capital investment. Information-based education, quality education, language training and K12 primary and secondary school extracurricular tutoring are witnessing iterative research results in “AI+Education”. What are the latest AI technologies in education? What is the current condition for their application? What’s the future for “AI+Education”?

On June 26-29, the 20th AIED (AI in Education) successfully ended in Chicago. Hundreds of top researchers and experts in AI+ education, computer science, psychology and other fields from all over the world gathered together at the conference to share and discuss the latest research results in AI, as well as the experience and challenges of AI application and AI products. First Insight, as an exclusively invited media in China, reported the whole process of AIED.

Organized by the 25-year-old International Association for AI Education (IAIED), AIED has over 1,000 members from more than 40 countries. They are experts and scholars in the fields of computer science, education and psychology. At the conference, Dr. Luis Von Ahn, founder of Duolingo, a famous language learning software; Candace Thille, professor of Stanford University and director of Amazon Learning Science Center; Nancy Law, professor of Hong Kong University; Jutta Treviranus, professor of OCAD University in Canada; and Richard Tong, chief architect of Squirrel AI Learning among others delivered keynote speeches and participated in round-table discussions. The co-chairs of the conference were Bruce McLauren, professor of Carnegie Mellon University in the US, and Rose Luckin, professor of education at University College London in UK.

AI is not a new thing in education. Although “AI+ Education” is a favorite target for capital investment, there are still doubts about its authenticity. Robby Robson, CEO of Eduworks and member of IEEE Standards Committee, pointed out that AI technology has been developing all the time and that all advanced machine behaviors are inseparable from AI, but there are five problems as follows:

1. Producers do not participate in research

2. Researchers are not engaged in production

3. Consumers do not know what to believe

4. Buyers do not know what to buy

5. Favorable technologies have not been applied to production.

Advertisement

In short, it is the separation between technology and industry that prevents AI from being applied. Parents of children in K12 primary and secondary schools are the true consumers of AI-related products, so it is especially crucial for them to understand and accept the important role of technology. Robby Robson also pointed out that the core of the application of AI technology is simple, that is, classification and decision. Only by understanding these two things can we find a way to connect machine learning with teaching.

At present, AI+Education mainly has two modes, namely the alternative and simplified traditional educational resources, and the personalized education. Alternative education emphasizes the application of technology to replace teachers in performing some tasks or strengthen teachers’ roles. Such products have developed rapidly, but their functions are relatively single. Personalized education using AI aims to complete the closed loop of “teaching, practicing and evaluating” and create customized learning method for each student, so that they will know which parts should be given up and which parts they should learn.

Intelligent adaptive system is a teaching method that uses algorithms to detect students’ learning paths and provide personalized and customized learning contents for students. Richard Tong, Chief architect of Squirrel AI Learning by Yixue Group said at the conference that, “Traditional school is like a bus which provides services in a simple and convenient manner, but it fails to take each student to the exact location they want to go to. Intelligent adaptive system is like a taxi that serves each student exclusively to meet their needs.”

Richard Tong, chief architect of Squirrel AI Learning, talked about the connection between human learning and machine learning as well as the way to completely change education and learning through AI. First, he talked about the reasons for applying AI in education and the significance of AI technology in replacing, enhancing, improving and redefining education. He then emphasized the continuous one-to-one individualized education for students and the change of the role of real teachers from instructors to psychological tutors for students, both of which are required by education revolution. He then introduced the reasons for the rapid development of Squirrel AI Learning from its business model, the experience of the application of AI, massive data in learning scenarios and strong computing power.

Squirrel AI Learning has created an adaptive learning engine driven by AI which integrates Mentality, Capability and Methodology (MCM), Probabilistic Knowledge State (PKS) model , Multimodal Integrated Behavioral Analysis (MIBA) using deep learning technology, Nano-Scale Knowledge Component (NKC) as well as other AI technologies like reinforced learning. This engine can promote the integration of AI and adaptive education and affect future education.

Richard Tong, chief architect of Squirrel AI Learning; Professor Rose Luckin, President of the Conference; and Ken Koedinger, Professor of Carnegie Mellon University and Chief Learning Scientist of Squirrel AI Learning, had a fireside dialogue. Richard Tong shared the success experience of Squirrel AI Learning in applying the above excellent researches into practice. The new model of “AI intelligent adaptive system + human teachers” makes for the maximum teaching effect and personalized learning experience for students. Professor Ken Koedinger of Carnegie Mellon University, a pioneer in cognitive science, shared his years’ experience of applying his research achievement in cognitive science and brain science to intelligent adaptive education products. Professor Ken Koedinger, chief learning officer of Squirrel AI Learning, discussed with Professor Rose Luckin about the technologies and values of Squirrel AI Learning’s intelligent adaptive teaching system and Professor Rose Luckin fully expressed his recognition. Richard Tong also held an industry forum discussion with Amy Baylor, program director of the National Nature Foundation of the United States, Brent Benson, online enterprise architect of Harvard Business School, and Kumar Garg, senior director of – Schmidt Futures, a public welfare organization set up by Eric Schmidt, co-founder of Google, to share the opportunities and challenges of their respective organizations in education science and technology. Richard talked about the importance of the cooperation between industry and academia. At present, there are four types of research fields and projects in education: learner model (to improve diagnosis and understanding of learning problems), domain model (to strengthen ontology and content), educational model (to enhance recommendation and channel optimization), and interactive model (to enhance interaction, feedback and communication). He also said that industry and academia can cooperate through various academic conferences and workshops, such as education-science and technology conferences, AI-machine learning conferences, industry conferences, workshops, tutorial samples, research papers and competitions.

On June 29, Squirrel AI Learning gathered IEEE AIS experts and professional audience of AIED to hold a workshop at DePaul University in the US to have a discuss on standardization of “AI+ Education”. Richard Tong, chief architect of Squirrel AI Learning; Robert Sottilare, senior director of Soar Technology; Robby Robson, member of IEEE Standards Committee and CEO of Eduworks; Xiangen Hu, professor of the University of Memphis; Avron Barr, Chairman of IEEE Learning Technology Standards Committee, together with experts, researchers and practitioners in AI education from Carnegie Mellon University, DePaul University as well as from Brazil, Cyprus and other countries, put forward their own opinions and had heated discussions about the standardization development of AI+ Education, AI’s integration with industry products, the application of intelligent adaptive education system, relevant algorithms and frameworks, etc.

For example, Robert Sottilare, senior director of Soar Technology, talked about the “Cases Study of Standards and Suggested Measures in Monitoring Tasks of Adaptive Teaching System”. AIS teaching areas include: cognition, problem solving and decision-making tasks, thinking; emotion, problem solving and decision-making, feeling; mental exercise, physical task; group, tasks involving two or more people, cooperated discussion. He finally asked a question to other scholars in the workshop: “The commonness in the instruction field can fulfill the following standards, namely the data exchanged in various AIS processes and the structural model in the domain. But what should the industry do to enable the AIS market to flourish? This is the next question to think about.”

As a leading enterprise in “AI+ Education”, Squirrel AI Learning is China’s first AI company to apply AI adaptive learning technology to primary and secondary education. It has successfully developed China’s first AI adaptive learning engine with complete independent intellectual property rights and advanced algorithms as the core. It uses more than ten algorithms and deep learning technologies and has developed many AI application technologies such as MCM capability value training, root cause reconstruction of knowledge map, nano-scale knowledge point splitting at the most basic level, relevance probability of non-relevance knowledge points, MIBA Multimodal Integrated Behavioral Analysis AI system, etc., the first of their kinds in the world.

As a Unicorn in China’s “AI+ Education” industry, Squirrel AI Learning has accumulated nearly 2 million student data in the past four years and has set up nearly 2,000 learning centers over 400 cities nationwide. Based on its deep understanding of education and continuous investment in technology, Squirrel AI Learning, a company with more learning data than others, will take personalized learning and improving learning efficiency as the direction of its future product development.

Advertisement

China’s AI technology is not inferior to that in foreign countries, but the development of AI+ Education lags far behind Europe and the United States since there are more than 50 AI education companies in Europe and the United Statesand 23 million users in the United States are using AI education products with an annual growth rate of between 30% and 40%. Given the rapid development of AI+education overseas and the huge development potential in China, Squirrel AI Learning, a leading enterprise in China, is actively tapping overseas resources and introducing and cooperating with global talents to develop the most advanced AI technology so as to make excellent intelligent adaptive learning products in line with the national conditions for Chinese students.

Squirrel AI Learning also plans to expand overseas to promote China’s advanced education platforms abroad, make products based on its existing educational knowledge maps, cooperate with overseas projects and build a curriculum system of mathematics that can be used worldwide.

At the conference, Professor Xiangen Hu of the University of Memphis shared his different views on the application of AI technology in education. He said: “AI technology should not be limited to the existing intelligent tutoring system – an easy-to-understand teaching system.” He proposed SIAIS (Self-Improvable Adaptive Instructional Systems) which consists of four parts. The key for this system is that learners and learning resources should match each other and they should make common progress. The core of “Self-Improvable Learning Resources” is educational games, functional teaching assistant systems and other dynamic contents. Therefore, it is necessary to define the variable but controllable modular components in the dynamic contents, combine them with big data and improve tutoring strategies through the interaction between learners and the dynamic resources.

As Nancy Law, a professor of the University of Hong Kong, said at the conference, “We should learn how to make full use of science and technology instead of being manipulated by it”. We should not turn technology into a gimmick, thus exacerbating arbitrary charge in the industry. Instead, we should use technology to provide better and fairer educational resources. Dr. Luis Von Ahn, founder of Duolingo and father of identifying code, said that although Duolingo is a technological product, his company always believes that education is the fairest way to change one’s fate, so they provide free services to customers and enhance learning entertainment with technology based on their understanding of learners’ changing learning psychology to help students become more motivated and improve users’ loyalty. Not just helping students solve learning problems and get high scores in exams, Squirrel AI Learning’s intelligence adaptive system aims to enable them to acquire good learning methods, get to know their advantages and disadvantages in thinking and personality. In this way, Squirrel AI Learning hopes to provide quality education for students.

Bruce McLauren, professor of Carnegie Mellon University, said in an interview that the new generation of users grow up together with technology. Students born in the era of Internet are eager for new technologies, new tools and new learning methods, which is both a challenge and an opportunity for enterprises. Only enterprises that provide quality education will be the winners.

SOURCE Squirrel AI Learning