First, personalized learning Although our brain structure is the same, everyone's way of thinking is completely different. For example, some people like to think and analyze problems with logic, while others are more accustomed to understanding things with perception. Many scientific studies have found that if students study in the way they naturally like, they will learn faster and their knowledge will leave a deeper impression in their brains. In addition, students will only store what they have learned in long-term memory when they think that the knowledge or skills they have learned are very important to them. Therefore, it is much better to learn what you are interested in than not to be interested.

Therefore, there is a consensus in the educational research field that teachers should first stimulate students' inspiration and interest before teaching children knowledge, and English is "stimulation". The highest level of education is not to instill knowledge or skills into students, but to explore students' own interests, so that students can actively study, think and innovate on this basis. This means that each student's subjects and themes are completely personalized and are chosen by the students themselves.

However, in the traditional school classroom, it is impossible to realize the complete personalization of teaching subjects and teaching methods. How can a class of forty or fifty people take care of everyone's hobbies? If this task is completely given to teachers, then a teacher can't take care of dozens or hundreds of students, which means that the teacher-student ratio needs to be greatly improved, and the labor cost of doing so is too high, perhaps only some aristocratic schools that charge sky-high tuition fees can do it.

Can machines share the work? The answer is yes. The rapid development of artificial intelligence makes the so-called adaptive learning possible. First of all, computers collect data about students' learning behavior and generate a lot of data about students' learning habits and preferences. Then through the systematic analysis of the data, the algorithm automatically adjusts the content of students' next study, recommends exercises suitable for students, and even changes the method of teaching knowledge. This process is ongoing. The more data, the more thoroughly the machine grasps the habits and preferences of students, the more accurate the matching of recommended contents and methods with students, and the learning efficiency will naturally improve.

Second, refined learning Many people have this experience when they go to school: it is easy to learn at first, but with the growth of grade, some subjects become more and more confused. Later, they had to rely on rote memorization and a lot of practice to cope with the exam and return it to the teacher immediately after the exam. In the end, it seems that they have learned nothing.

Why is this happening? The reason is that our learning process is too rough, and we are in a hurry to learn the next concept before we fully understand it. A lot of knowledge, especially scientific knowledge, is interrelated. If the former concept is not fully understood, the next one will be a little confused, and the next one will be completely confused. At this time, many people may feel that they are "not suitable for this subject", start to hate this subject, and even lose confidence in themselves. In fact, many times, this situation has little to do with one's own intelligence and ability, just because one link in the knowledge chain is not mastered well, so the whole chain is broken.

Really effective learning should be refined, just like the apprenticeship system of European craftsmen hundreds of years ago: every step of a craft must be practiced to perfection before the next step can be started. An apprentice, without more than ten years' practice, can't make something that sells at a good price like what the master made. Nowadays, it is difficult to find such detailed study. With the explosion of knowledge and the flood of information, our time has become "fragmented". Everyone has no patience to study in depth, and many things rush into the next step just by learning a little fur.

Although not all knowledge and skills need to be studied seriously, meticulous study can never be ignored in K 12 education to cultivate children's learning methods and habits. In order to cultivate students' rigorous thinking habits and meticulous logic, they need to have a comprehensive and profound understanding of at least one subject. This undoubtedly requires a lot of teaching. In a class of forty or fifty students, the teacher can only teach according to the average progress of everyone, but no one in the class may have the same learning progress as the teacher. This kind of teaching will certainly not enable students to achieve refined learning.

Artificial intelligence can change this situation. The progress of students' learning is not determined by the teacher, but is debugged in real time by observing the machine of students' learning at all times. Artificial intelligence algorithm infers what knowledge points students have not mastered according to the performance of students' practice, and then makes up for the defects in learning by strengthening practice, reviewing concepts and giving more examples until the knowledge system of the whole chain is fully mastered. The advantage of this method is that on the one hand, it prevents students from jumping to the next step before mastering the necessary knowledge, on the other hand, it also saves the time of doing useless exercises and greatly improves the learning efficiency.