iTutorGroup Shen Peihong talks about AI empowerment education: the algorithm is powerful enough "1 to 3" will learn 20% faster than "1 to 1" students

[TechWeb] on April 15, Shen Pei-hung, chief data officer of iTutorGroup, an international online education brand, attended the 2012 Annual artificial Intelligence Industry Summit of FUS Orion, and introduced the practices of iTutorGroup on the application of AI in education.


Picture 1

ITutorGroup Shen Pei-hung

For how to apply AI and big data to education, in the view of Shen Pei-hong, a correct AI application in the online education scene is definitely not a substitute for a real person, but an auxiliary real person. The greatest value of AI in educational applications is the "to substitute his or her hand and hand to make up his ear". It is the method of online learning that students do not have to go to the other end of the earth to follow a famous teacher. It is the help of the system, which makes the teacher more intelligent, and has a very clear grasp of all the interaction and the students' reaction in the class.

Macro and micro operations of AI empowerment education

The application of AI in education makes the individualized teaching according to students' aptitude to be implemented on a large scale. Shen Pei-hung said that to achieve personalized education, AI can play a role at both the macro and micro levels.

“The most important thing at the macro level is how to determine the best learning path for each student through the algorithm,” Shen Peihong said. “When the students’ problems reflected in the learning process are cut into small enough, the maintenance is cut enough. When there are many, there are no phenomena observed by any two students or the results obtained are the same."

Picture 2

How can data and AI technology help to create a personalized learning path for students? In the iTutorGroup, DCGS is the key to the entire system. In the iTutorGroup media open day in February this year, Shen Peihong once told TechWeb that the DCGS system did a feature analysis of more than 30,000 textbooks, more than 20,000 teachers and 180 million person-times of learning processes, labeling different attributes. With the help of the label, the system independently helps the students to customize the course according to the interest, background, ability and other special quality, match the teacher, and achieve "personalized" teaching.

After the macro personalization, the next step is the micro-level AI empowerment.

Shen Peihong believes that at the micro level, a correct AI application is definitely not to replace the real person in the online education scene, but to assist the real person. The value of the real person is at the micro level. That is to say, the algorithm determines the learning trajectory of each person. When it is time to go to class, a real teacher can find the best way for students in this class to absorb the teaching content through his teaching experience and teaching skills.

The role of teacher includes preaching, teaching and solving, preaching and teaching is the transmission of content, this is more likely to rely on the system to replace artificial, and no artificial intelligence to replace the teacher's position.

In addition, through real-time big data monitoring, master the effect of each class to learn whether students have encountered obstacles in learning. The information collected by the system is sufficiently deep enough to allow the assistants or customer service staff to make advance-looking care before the students are tired of learning and dissatisfaction accumulated in the outbreak point, and eliminate the problem invisible.

1-3 is more efficient than 1-1 when the algorithm is strong enough

Shen Peihong introduced that iTutorGroup can provide students with one-to-three, one-to-four small classes for individualization and teaching goals in front of thousands of people, and it is not necessary to be forced to make one-on-one to satisfy personalization.


The online education 1 to 1 model is costly, and the 1 to n small class model is helpful for maximizing teacher resources, competing incentives among students, and reducing learning costs. It can provide a one-to-one to one-to-one personalized teaching mode, which is more cost-effective. In Shen Peihong's view, it is also possible to achieve iTutorGroup's three goals of “learning, learning, and learning” in education.

iTutorGroup 沈沛鸿 talk AI Fu algorithm enough powerful 1 pair 3 compare 1 pair 1 student learning fast 20%

Read More Stories

© , New View Book , Powered by UIHU