Recommendation
Big models are not popular in basic education, and data problems have become a key bottleneck.
Core content:
1. The application dilemma of DeepSeek big models after local deployment
2. The three major problems faced by the digitization of basic education data
3. A feasible solution to promote the digitization of paper materials
Yang Fangxian
Founder of 53AI/Most Valuable Expert of Tencent Cloud (TVP)
After DeepSeek announced its open source, on January 30, 2025, the courseware teacher wrote a public account, which was read 82,000 times, forwarded by more than 6,000 people, and reposted by 13 public accounts. " The significance of DeepSeek local deployment for primary and secondary education: bringing educational data to life at low cost, high efficiency, and security "
After the long Spring Festival holiday, many schools and government agencies announced that they had deployed the DeepSeek big model locally. However, after the release, it was often "much ado about nothing" and there was no follow-up in terms of substantive application. The reason is simple: without data support, the big model is like water without a source . Even if the technology is advanced, if the data of basic education still remains in paper form, AI will not be able to play a role. This is the fundamental reason why many projects "have no follow-up after more than a month of deployment."
The core of artificial intelligence lies in the accumulation and analysis of data. Through the training of massive data, AI can identify patterns, optimize algorithms, and provide users with personalized solutions. In basic education, this seemingly simple "data accumulation" has become the biggest obstacle. Why? The answer is: the data of basic education cannot be digitized at low cost .
In ordinary primary and secondary schools, whether it is student homework, textbooks, or classroom exercises, most of them are in paper form . These paper materials cannot be directly read and processed by computers, and naturally cannot provide the "nutrients" needed for AI. In other words, no matter how smart the AI "brain" is, it can only "run in vain" without data to feed it.
Why is it difficult to digitize data on basic education? We can analyze it from the following aspects:
1. Equipment restrictions. In university education, students generally use mobile phones, tablets or laptops. Teachers can assign homework and collect data through online platforms, and the entire teaching process is highly digitized. Basic education is completely different. Primary school students are explicitly prohibited from bringing mobile phones into the classroom, and there are even policy interpretations that "mobile phones should not appear in the classroom." The use of tablets is also strictly restricted for reasons such as vision protection and preventing addiction to games. Without electronic devices, data cannot be collected in real time.2. Policy barriers. Education policy clearly stipulates: "Mobile phones are not allowed in the classroom unless necessary for teaching." This is not only to prevent students from becoming addicted to electronic products, but also out of consideration for classroom order and student health. However, this regulation has invisibly blocked the path to data digitization, making the application of AI technology in basic education difficult.3. Insufficient technology and hardware. Even without a mobile phone or tablet, paper materials can be digitized through scanning equipment. But the reality is that the popularity of high-speed scanners or high-speed cameras in basic education is far from enough. Many schools lack funding or technical support to quickly digitize a large number of books, homework and student exercises.It is the combination of these factors that has caused the data in basic education to always remain on paper and unable to enter computers, making the application of AI out of the question.
The way out: promoting the digitization of paper materials
In order for AI to play a role in basic education, the first task is to solve the problem of data digitization. Here are several feasible solutions:
1. Introduce high-speed scanning equipment. Schools can be equipped with high-speed scanners or high-speed scanners to quickly convert paper books, homework and students' completion status into electronic formats. These devices are not only easy to operate, but also can greatly improve data collection efficiency and provide rich materials for AI. This view is elaborated in detail by the courseware teacher in this article " In the era of artificial intelligence, high-speed scanners should become the standard equipment of "Banbantong" equipment "2. With the help of multimodal technology. Based on the electronicization of data, large models and multimodal terminal design can be used to input text, images and even sound data into computers for in-depth analysis. Multimodal technology can process multiple data types, allowing AI to understand students' learning more comprehensively.3. Develop a dedicated education platform. Within the scope permitted by policy, design a dedicated education platform for teachers and students to complete teaching tasks using electronic devices in a controlled environment. This platform can integrate data collection and analysis functions, which not only meets policy requirements but also paves the way for AI applications.4. Reform the distribution model of homework books. With the advent of the era of artificial intelligence, all industries should think about themselves and how they should change. Unified printing of homework books does not conform to the concept of precise teaching and personalized teaching. Don’t customize the printing of homework exercises like pre-made dishes, but deliver them in real time according to the teaching progress and students’ mastery of the situation. Of course, this is too idealistic. Here are the links to related articles in this official account: " Homework printing factories also need to undergo digital transformation, otherwise they will be eliminated by the times " and " Move the printing factory into the teacher's office: a vision of the future of personalized homework "Despite the challenges, the application of artificial intelligence in basic education is not unsolvable. In the future, we need to work together in both technology and policy:
1. Policy adjustment: Under the premise of ensuring students’ health and classroom order, moderately relax restrictions on the use of electronic devices and explore safe and controllable digital teaching models.2. Technological innovation: Increase investment in educational technology and promote efficient data collection tools, such as high-speed scanners and multimodal terminals, to make the digitization of paper materials simpler and more common.3. Training support: Provide technical training for teachers and students to improve their acceptance and ability to use AI tools and digital devices.The essence of artificial intelligence is the accumulation of data, and the biggest problem with AI applications in basic education is that data cannot be digitized. Only by overcoming this obstacle and converting paper materials into computer-processable data through technological innovation and policy support can AI truly enter the classroom and inject new vitality into education reform.
