Shen ran Cai Jing original

Author | Tang Yahua

Edit | Dawn

The annual college entrance examination is coming again. In Weibo, there were 90.594 million readings and 24,000 discussions on the topic # How to brush the topic after a brief encounter #. A real and cruel problem is that the tactics of questioning the sea may also be equivalent to ineffective brushing. Can I really brush the questions?

There is no shortcut to education, only a hundred trials. In this process, technology cannot change the essence, but it can improve efficiency.

Some people say that wrong questions are the DNA of learning, which not only exposes the weak points of students' knowledge structure, but also further infers the lack of ability behind weak knowledge. Then, the AI algorithm can make people want to brush Tik Tok and play games. Will it fascinate us as much as brushing short videos?

There is such a product that is dedicated to the personalization of learning paths. Generally speaking, it is the place where you can't learn, check for leaks and fill gaps, and bid farewell to the vast sea of questions.

Intelligent teaching assistant, also called artificial intelligence adaptive education, is called the ultimate "weapon" of personalized education. The way to brush the sense of existence is to let students create their own AI personal education instead of asking questions about the sea.

A few years ago, AI+ education once became a hot spot, and technology was considered as a magic weapon to solve the scarcity of educational resources and personalized education. However, the development of reality is slower than ideal. In recent years, I don't know whether AI will retreat if it doesn't advance, or whether capital and market have forgotten this once "pearl". The trend of online education has changed suddenly, and the focus has shifted to the live class of double teachers in large classes.

Nowadays, the market direction is uncertain, and some enterprises in the track have begun to regain their intelligent teaching AIDS. Today, let's take a look at the intelligent teaching AIDS that can screen and evaluate students' problems through AI, and then recommend personalized learning programs. Is it a gimmick or a real kung fu? Has the industry developed to today's level? To what extent can such technology help middle school students prepare for the exam?

"Old wine in new bottles", AI-assisted learning is coming again? Zuo Yan, a junior three student, has always been a bully in the university. He thinks that he has brushed more math problems than ordinary students, but in two important exams last year, he encountered several problems that he couldn't do, and the final result was surprising. Only then did he find that although he had brushed a lot of questions, there were still blind spots in knowledge, and some low-quality test papers and question banks were often repeated, wasting time.

In addition, Zuo Yan summed up the reasons. Although he wrote many wrong questions, he didn't find the commonness of the wrong questions. After changing the form, he might not do the same knowledge point. "In fact, I didn't understand the nature of knowledge points. I don't understand how the formula came from and how it was derived. In fact, all the questions are about one knowledge point or several knowledge points. If I can solve the problem, I can look at it from a macro perspective and move closer to the knowledge point step by step. "

Later, he tried to use some intelligent teaching systems, including an intelligent question bank. He found: "The system can help me trace back to the source, find the essential source of mistakes, and clearly explain how this knowledge point is applied to various topics. Understand these essential laws, you will have a global vision, and finally doing the problem is like knowing how to solve the cow. "

In the market, companies such as ape question bank and squirrel AI intelligent adaptation have such products. Recently, Netease announced to upgrade the intelligent teaching assistant system, claiming that it can evaluate students' ability, generate learning paths, improve orientation, and check for missing parts.

With the approach of the senior high school entrance examination, Zuo Yan suddenly didn't know how to choose, and the blind tactics of asking questions about the sea were no longer applicable. Is it reliable to give precious time to a system?

Deep combustion experience shows that this type of system usually includes evaluation, finding problems and recommending corresponding exercises. For example, Netease's intelligent teaching assistant system is embedded in the excellent course software. Its "ability evaluation" means that students can choose a certain section of a certain learning cycle, such as "intersection line and parallel line", and then the system pops up five questions. After making a choice, the result of mastering the situation of a certain topic will appear.

The system shows, "The score is D, and there are 6 knowledge points in this question. You have passed 0 knowledge points, but there are still some knowledge points to pass, such as vertex angle, adjacent complementary angle and vertical line. AI private education suggests that you continue to study. " Then the system will generate a learning path. According to the instructions of the system, there are recorded real teachers to explain the knowledge points, and there will be corresponding exercises after the video is completed, and so on.

In fact, this is not a new concept in online education. Since 20 15, companies specializing in AI+ education have mushroomed, which is also called "adaptive" or "intellectual adaptation". In the field of internet, which is good at coinage, "adaptation" is a marketing gimmick. A more decent way of saying this model is personalized teaching. Generally speaking, it is algorithm recommendation in the field of online education. Short videos and e-commerce let you see what you like, and learning to do problems makes you unable to learn anything.

It is worth noting that this is only a link in the application of AI in the field of education. At present, the products on the market cover the most peripheral photo search products, such as homework help's and Little Ape's photo search products, as well as "spoken English assessment". On behalf of enterprises, they are fluent in English, box fish and so on. The adaptive learning mentioned above acts on the core learning link.

Just look at the adaptation in the field of K 12. The difference among different schools is that ape tutoring is a simple adaptation item bank product. Squirrel AI intelligence matching mainly focuses on after-school tutoring, consolidation exercises, and leak detection. 90% of the students are offline, and there are also professional teachers to guide them in class.

Netease's well-known system emphasizes the combination of teachers and intelligent systems in content and the division of knowledge modules. They combine the intelligent teaching assistant system with lecturers and tutors and use it in class and after class respectively. At the same time, Netease claims that its online junior high school mathematics is divided into more than 4,000 knowledge modules, and each module only needs 5 questions on average, and the weak points can be found in 3 minutes. 1 second gives the learning path.

AI is better than one-on-one tutor? The Internet in China is a technology tree. As described in "The Essence of Technology", "Times create technology, and technology also creates times."

From around 2000, with the popularization of computers, Hujiang Online School came into being, with the earliest online courses, which can be called the first internetization of the education industry. Around 20 10, many people began to come into contact with the open classes of universities around the world that they could not reach before.

With the rise of mobile Internet, with mobile phones and 3G and 4G networks, in 20 14 -20 15 years, the education industry was accompanied by the first wave of mobile Internet boom, and tutoring O2O became popular. Then, live classes and online one-on-one teaching appeared. With the improvement of the live broadcast system, large classes also appeared.

At this time, entrepreneurs have realized that the education field that needs personalization and efficiency improvement urgently needs the blessing of technology.

The concept of individualized adaptive education was put forward in the United States in the 1990s. The core idea is to split the knowledge system, evaluate the students, and finally generate a large-scale teaching product of "teaching students in accordance with their aptitude". The famous Khan Academy in the United States also announced the transformation of AI adaptive education mode on 20 17.

The development of K 12 adaptive education track started late. On 20 15, squirrel AI intellectual adaptation (preschool education) was established, and an adaptive learning system was introduced. From 20 17, the money in the education industry began to chase after AI+ education. This year, Squirrel AI Zhi Zhi completed the Pre-A round of financing; Extreme big data and intellectual education, which focus on the research and development of educational intelligent systems, respectively received 1 100 million yuan series B financing and 200 million yuan series B+ financing; Hujiang launched "Uni Intelligent Learning System" and advocated "tailor-made" personalized learning.

Nowadays, the "AI+ Education" track has covered the whole industrial chain of "learning, practice, reform, test and evaluation", and a series of functions such as oral evaluation, composition correction, homework assignment and adaptive learning are trying to create an intelligent and efficient learning environment for students.

What is the level of adaptive education? This is nothing new.

Zhou Wei, co-founder of squirrel AI intelligent adaptive, told Shen Zhan: "The current AI adaptive can solve three problems, namely personalized learning, educational equity and public welfare".

Personalization is easy to understand. Educational equity means that students in areas with insufficient educational resources can enjoy high-quality resources through this technology, while public welfare means price. A system can be used by an infinite number of users, and the cost of this product can be reduced under the marginal effect.

Regarding the product effect, Zhou Wei pointed out: "In the inquiry and trapping, the system has long-term data tracking for students, and students' judgments may be more accurate than teachers; Second, the planning ability of the system is strong, because there is enough data, which can be adjusted in real time like map navigation; Third, because high-quality teachers can be concentrated, the system is more stable in the transmission of knowledge points. "

Curiously, what is Netease doing at this time?

In fact, as early as 20 19, Netease has already made a smart pen, which has accumulated a lot of experience for the intelligent teaching assistant system. But more importantly, Zhou Feng, CEO of Netease from a technical background, has always had a product dream and is unwilling to give up using technology to form products from educational standardized content.

Zhou Feng not only believes in technology, but also advocates individualized teaching. In Zhou Feng's student career, the physics teacher had the greatest influence on him. "I learned physics very well in middle school. The teacher took me to the office to teach one-on-one at home. Finally, I participated in the physics competition and won the seventh place in Jiangsu Province. " Zhou Feng said.

In reality, not every student has such an opportunity, so Zhou Feng set a clear goal for Netease, "Let every student achieve the effect of one-on-one teaching by experienced teachers through online classes".

In his understanding, a new generation of teachers should know technology. He believes that the effect of personalized teaching in the industry for so many years is not obvious because the combination of teachers and technology is not close enough. Netease has a way to put teachers and technicians together. "No matter what the two groups argue, you should work together. Noisy, let's find a way together. The actual product will not leave the course, and the teacher will use it in class. "

Is the intelligent teaching assistant system reliable? There is no doubt that AI is a powerful tool, but in the field of AI+ education, only practitioners who really understand education can create products that conform to the laws of education.

Among online education enterprises, homework help, Ape Tutor and Netease Youdao are actually the founders of technical schools and all have a product dream. Homework help originated from Baidu, and the founding team of Ape Tutoring came from Netease.

Zhou Feng, CEO of Netease Youdao, said with a smile that Youdao and Netease are both product companies, and what they are best at is how to find solutions to many people's problems, instead of solving the problems of a few users like manual workshops. "We used to say that education should be industrialized, and people don't like to hear it. Most people use products from industrial production lines, but they say they like handcraft. In fact, you really dare not drive a manual car. " Therefore, on the basis of educational technology, the industry needs to improve efficiency and give consideration to user value in a better way.

Obviously, the intelligent teaching assistant system is a product that Zhou Feng thinks can play this role.

Practicing public welfare through business is the best way for entrepreneurs to assume social responsibility. Using technology to enlarge educational resources, improve learning efficiency and reduce the cost of students' personalized learning is equivalent to social welfare to some extent. In the future, with a firm vision and mission under the conscious supervision of entrepreneurs and policy science, AI+ education may be able to draw a beautiful map of the new world.

However, the concept and slogan are hot, and AI+ education has been tossing for several years. Apart from the noise of capital and the agitation of entrepreneurs, the legendary subversion does not seem to have come, and people's worries and doubts about it have never stopped.

Jiang Yonghong, vice chairman of China Education Improvement Society, told Shen Ran that photo search, oral error correction, composition correction and self-adaptation are all applications of artificial intelligence in the field of education by using speech recognition, character recognition and picture recognition.

In his view, the value of photo search is limited. "Since students can't do problems, it's useless to give them answers, because the writing of answers is deductive, and solving problems depends on analysis. We may use various assumptions, reasoning and judgments to exclude and get a correct result. Students who find the answer can't understand the problem-solving ideas, which is exactly the most valuable part of students' growth needs. "

He also mentioned that AI adaptation has certain value from the perspective of knowledge points. After all, there are many knowledge points in middle school, and most students can't do the questions because they don't understand them thoroughly enough.

However, the shortcomings of AI are also obvious: "First of all, AI personalized training must be based on students' active learning. If a student has strong self-control and patience and browses all the recommended information, he can achieve the expected learning effect. If he can't stick to it, it is meaningless to recommend it accurately. " Jiang Yonghong said.

In addition, in disciplines, AI adaptation is more closed knowledge, for example, it is better used in mathematics and physics, while knowledge points such as Chinese are not so regular, and the parts involving numerical values are also difficult to solve.

For college students, although their grades are only temporary gains and losses, the senior high school entrance examination may only be a life experience, but this examination will also directly determine what school they can go to, what classmates they contact, what learning atmosphere they are in and what values they form, which is also the basis for their future foothold between heaven and earth.

Hemp rope is a detailed break and sprint designed for the college entrance examination. In the end, the more personalized learning paths are needed. Today's AI may not be able to shoulder the heavy responsibility of middle school students' education in the whole discipline and learning process, but it may play some roles in relatively standardized disciplines, find the crux of the problem for each student, and help them sweep mines and pits and move forward efficiently.

* The title map is from vision china, and the pictures in the text are all from Pexels. At the request of the interviewee, Zuo Yan is a pseudonym.