Citation Format
Sun Faqin, Xu Nuo. Reflections on Artificial Intelligence Teaching Content in Primary and Secondary Schools from the Perspective of Core Literacy [J]. China Educational Informatization,2023,29(8):113-120. DOI: 10.3969/j.issn.1673-8454.2023.08.013
Research on the Digital Development of Basic Education
Reflections on Artificial Intelligence Teaching Content in Primary and Secondary Schools from the Perspective of Core Literacy
Sun Faqin Xu Nuo
Abstract: The new round of technological and industrial revolutions has led to significant changes in talent demand, posing new requirements for education. Establishing artificial intelligence-related courses in primary and secondary schools has become a trend. However, unclear understanding of the goals of AI courses, misconceptions about course content, and a lack of quality teaching materials and faculty have severely hindered the popularization and promotion of AI courses in primary and secondary schools. Centered on the core competencies of the AI era (scientific literacy, new digital literacy, and new humanistic literacy), AI courses in primary and secondary schools can be organized around five aspects: how AI perceives the world, how AI represents the world, how it learns from data, how to collaborate with AI, and how to evaluate AI applications. Suggestions for course implementation are proposed at three levels: experiencing intelligent life, understanding intelligent applications, and participating in intelligent innovation.Keywords: Artificial Intelligence; Teaching Content; Primary and Secondary SchoolsClassification Number: G434Document Identification Code: AArticle Number: 1673-8454（2023）08-0113-08
Author Information: Sun Faqin, Associate Professor and Master’s Supervisor at the School of Journalism and Communication, Yangzhou University, PhD (Yangzhou, Jiangsu 225009); Xu Nuo, Teacher at Suzhou University Experimental School (Suzhou, Jiangsu 215133)
Funding Project: 2017 National Social Science Fund (Education) project “Research on Fragmented Reading Learning Behavior and Cognitive Depth Enhancement” (Project No.: BCA170084) Phase Results1. Background of AI Course Implementation in Primary and Secondary Schools
The Gartner report predicts that by 2045, 50% of today’s jobs may be replaced by Artificial Intelligence (AI).[1] AI will increase productivity in many jobs, eliminate millions of middle and low-level positions, while also creating more high-skill, high-management new positions.[2] Many past significant innovations have caused temporary unemployment for some workers, such as the invention of the steam engine leading to job losses for some hand-weavers, and the “Internet+” causing job losses for some bank employees. So, will today’s K12 students face unemployment right after graduation when they enter the workforce? Therefore, to adapt to this era, K12 schools must offer relevant AI courses.
As the most important transformative force in modern society, AI is reshaping our lives. AI has become a new driving force, just like the internal combustion engine, electricity, and network technology, providing tremendous power for our production and life. Today, the successful applications of AI in Amazon’s recommendation engine, Google Translate, and Apple’s Siri, AlphaGo defeating humans in Go, and IBM Watson providing optimal treatment plans for current cancer treatments, have achieved great success in areas such as autonomous driving, medical imaging diagnosis, language translation, and speech recognition. As a foundational technology, AI is increasingly valued and has penetrated various industries, successfully assisting many traditional industries in achieving leapfrog upgrades. In contrast, the application of AI in education has limited impact. Today’s K12 students are unaware of how AI works and the opportunities and challenges it will bring. Therefore, to keep pace with innovation, students must understand this technology from a young age.
Since the concept of artificial intelligence first entered the “13th Five-Year Plan for National Economic and Social Development of the People’s Republic of China” in March 2016, various ministries have densely published documents regarding AI, fully reflecting the country’s top-level design and strategic deployment for the AI industry. From the State Council to the Ministry of Education, between January 2017 and February 2019, five documents directly related to the introduction of AI into primary and secondary schools were successively released, fully reflecting the country’s emphasis on offering AI courses in primary and secondary schools. The key work in 2019 was to “promote the establishment of AI-related courses in primary and secondary schools and gradually promote programming education.” Therefore, there is an urgent need to build suitable AI education-related courses for primary and secondary schools.
2. Current Situation Analysis
In April 2021, I conducted a survey on the implementation of AI courses in 15 schools in my city. The surveyed schools included 5 high schools, 4 junior high schools, 4 elementary schools, and 2 integrated schools (covering compulsory education from grades 1-9). The survey content included whether AI courses were offered and the reasons, the timing of implementation, the content offered, the experimental platforms used, the sources of teaching materials, and the status of faculty.
(a) Few Courses Offered
The survey found that among the 15 schools, only one high school offered a module on “Introduction to Artificial Intelligence” as part of its elective information technology curriculum. This school implemented the elective information technology curriculum starting in the fall of 2018, requiring all students to choose one from five elective modules (drones, basic Python, 3D printing, APP Inventor, Introduction to Artificial Intelligence). The other 14 schools did not offer any courses related to AI or courses that included AI modules.
(b) Varied Teaching Content
During the survey, most teachers in schools could not distinguish between STEM education, robotics education, programming education, and AI education. They believed that AI education was synonymous with robotics education and STEM education. Many schools claimed to have already offered AI courses, but upon further investigation, it was found that they equated AI education with building robots and simple block-based coding similar to Scratch. Additionally, the exaggerated promotion by social tutoring institutions misled schools and teachers, such as one institution positioning its robot platform as an “innovation platform for cultivating future AI scientists.”
(c) Quality of Teaching Materials Varies
Currently, most teaching materials used for AI courses are self-compiled handouts, many of which are compiled by information technology teachers from simple cases collected online, lacking coherence and continuity. In recent years, some AI textbooks have been published, such as “Introduction to Artificial Intelligence (High School Edition)” edited by Tang Xiaowu and Chen Yukun, published by East China Normal University Press, which has a very reasonable content arrangement. However, my research on this textbook found that it is filled with numerous data formulas (many related to higher mathematics and signal and system courses), making it quite challenging to implement in ordinary high schools where information technology is not included in the total score for college entrance examinations.
(d) Diverse Experimental Platforms
From the current schools offering AI courses, most still use Python-based experimental platforms. Many schools also use various platforms provided by robot manufacturers, such as Lego, Makeblock, and others, which have developed rapidly in recent years. These manufacturers not only provide hardware facilities but also offer supporting teaching materials for implementing courses on their platforms. There are many examples of confusing AI with STEM education and robotics education.
(e) Weak Faculty Strength
According to the survey, the relevant faculty mainly comes from information technology teachers, graduate students from partner universities, and some lecturers from robot manufacturers. Due to the heavy course load of information technology teachers and outdated knowledge structures, the quality of graduate students from partner universities varies greatly and they have high turnover rates, while the knowledge of manufacturer lecturers is limited and their service time is short. Currently, the implementation of AI courses in most schools remains quite challenging.
The unclear understanding of AI course objectives, misconceptions about course content, the lack of quality teaching materials, and the weakness of faculty have led to a very low implementation rate of AI courses in primary and secondary schools. Therefore, clarifying the overall objectives of AI courses in primary and secondary schools, establishing suitable course content for learning at this stage, and designing implementable course plans are fundamental ways to solve the above problems.
3. Core Competencies in the AI Era
The future AI era will be one of human-machine collaboration and coexistence.[3] What core competencies should the emerging generation possess in this context?
First is scientific literacy. Scientific literacy is the ability to interact with the physical world, providing the principles of disciplines, engineering technology, and programming skills needed in the AI era, enabling people to better understand machine behavior. Second is digital literacy. With the rapid expansion of information from the internet and the Internet of Things in recent years, information has begun to swell rapidly, directly from the physical world, making it impossible for humans to process this information directly, forming a new information space.[4] Digital literacy provides the ability to interact with the information world. Finally, and most importantly, is humanistic literacy. The importance of humans in the AI era is reflected in curiosity, creativity, responsibility, collaboration, critical innovative thinking, and social skills, as shown in Figure 1.
Figure 1 Core Competencies in the AI Era
(a) Scientific Literacy: The Ability to Interact with the Physical World
The current rise of AI is rooted in breakthroughs in deep learning theory and the background of big data cloud computing. However, while deep learning allows AI to conquer Go, it cannot endow it with the reasoning abilities unique to humans. Recent research by the DeepMind team on AI solving math problems found that AI performed poorly, answering only 14 out of 40 high school-level math problems correctly, with an accuracy rate of only 35%.[5] AI excels in pattern matching, machine translation, and reinforcement learning, while humans learn primarily through inference, learning, and using theorems, axioms, and rules of symbolic operations. Humans can easily extrapolate things beyond their existing experiences, while AI cannot. Therefore, current AI is more of human intelligence. Only humans can learn scientific knowledge, master the scientific research process and methods, and understand the impact of science and technology on society and individuals.
Today’s primary and secondary school students are “digital natives” who grow up in a digital environment and inherently possess the “genes” to be familiar with various digital devices. Although they know how to use these devices, they may not understand how they work, even at the most basic level. Just as we needed to understand the basic structure of an engine 100 years ago, understanding the basic principles behind engines is essential to maximize their utility and productivity to serve the early industrial revolution.
(b) New Digital Literacy: The Ability to Interact with the Information World
Scholars proposed the concept of digital literacy as early as 1997, defining it as the ability to acquire, understand, and integrate digital information.[6] Other scholars believe that digital literacy is the ability to read, understand, create, and communicate data in the form of information.[7] New digital literacy refers to the new capability system that “digital natives” should possess in the post-digital era to adapt to future digital life, including data thinking, computational thinking, and programming skills. Data thinking supports macro system design, computational thinking aids in meso process design, and programming skills ensure micro implementation.
New digital literacy includes sensitivity to data, data collection ability, data analysis and processing ability, decision-making ability based on data, and the ability to evaluate the results of data usage, as shown in Figure 2. Data sensitivity, as a foundational ability, requires a comprehensive grasp of the digital essence of things, perceiving changes in the state of things through data variations. On this basis, data collection ability involves knowing where to obtain the necessary data and how to acquire it. Further enhancement requires data analysis and processing ability, which involves using various information processing tools to manipulate and process data. With these abilities, we can make data-driven decisions and create value. Finally, data evaluation ability refers to the ability to objectively assess the reliability of data and the reliability of analysis results, further improving the quality and effectiveness of data applications. Thus, these abilities form a gradually deepening and expanding hierarchical structure, constituting an important part of new digital literacy.
Figure 2 Levels of New Digital Literacy
The application of various new technologies in our lives has led to a dramatic increase in data. New digital literacy can help us grasp the digital essence of things, understand the true meaning of data, find value, and assist in decision-making.
(c) New Humanistic Literacy: The Ability to Interact with the Human World
In the future, AI will easily replace repetitive and monotonous tasks, but it cannot replace the most subtle aspects of humanity. Such abilities include comprehensive analysis and decision-making, emotional communication and empathy, creative thinking and imagination, teamwork and interaction. These abilities are the core that distinguishes humans from AI. To differentiate from traditional humanistic literacy, this paper collectively refers to these abilities as new humanistic literacy.
1. Comprehensive Analysis and Decision-Making Ability
This requires cross-disciplinary thinking, breaking existing methods, broad referencing, returning to essence, finding principles beneath rules and trends, and discovering effective solutions through unconventional approaches. Current AI can significantly assist law firms in identifying relevant documents in legal cases, but ultimately, a human judge is still needed to make rulings; similarly, in the medical field, no matter how advanced medical equipment is, it cannot replace clinical physicians in making medical diagnoses.
2. Emotional Communication and Empathy
Caring, compassion, and emotional connections are unique human abilities, representing the greatest and most enduring advantage of humans over machines. The essence of machines lacking a soul will always mean that their interactions with humans are “cold”; no matter how lifelike technology can make them appear, machines can never empathize with humans.
3. Creative Thinking and Imagination
Imagination is a quality unique to humans. Currently, AI operates by utilizing existing data and performing logical reasoning based on parameters we provide; imagination and dreaming cannot be achieved through programming. Curiosity and imagination drive innovation and are key to problem-solving, which is precisely what humans excel at. Throughout history, curiosity and imagination have propelled scientific advancement.
4. Teamwork and Interaction
Teamwork is a special ability of biological groups, where team members collaborate, communicate effectively, anticipate and meet each other’s needs, and inspire confidence, forming coordinated collective actions. Through this ability, humans can enhance each other’s complementary advantages in collaboration with computers, combining human leadership, teamwork, creativity, and social skills with the speed, scalability, and computational power of machines to create collaborative intelligence, ultimately supplementing and enhancing human capabilities.
4. Suggestions for Course Content
Based on the core competencies that the emerging generation should possess in the AI era, and referring to the framework guidelines proposed by the AI4K12 (AI for K-12 Initiative), the content of AI courses in primary and secondary schools needs to inform students how AI interacts with the physical world, how it interacts with the information world to generate intelligence, and how humans collaborate with AI to ensure that AI systems operate normally, ethically, and safely.
(a) How to Perceive the World
First, students should understand how computers perceive the world. Humans can observe the world with their eyes and listen to sounds with their ears, while computers recognize the world through various sensors. Cameras and ultrasonic sensors can be seen as the eyes of computers; through cameras, computers can recognize various images, such as facial recognition and motion detection, while ultrasonic sensors can identify distances. The current capabilities of various sensors are one of the reasons for AI’s success, so it is essential for students to understand the working principles and usage of relevant sensors as the first step in learning AI. By understanding relevant sensors, students can identify the usage scenarios and limitations of various sensors to avoid scientific errors in future innovative activities. Therefore, the AI curriculum in primary and secondary schools should include (but not be limited to) image/color sensors, ultrasonic sensors, touch sensors, sound sensors, temperature and humidity sensors, light sensors, gyroscopes, GPS, pressure sensors, etc.
(b) How to Represent the World
Computers use abstract data models to represent the world and perform calculations and reasoning with them. Representation is one of the fundamental characteristics of intelligence, whether human or artificial. Therefore, students should first understand abstract concepts, such as using boundaries on a map to represent an area, or a table to represent a board game, as shown in Figure 3, and then understand how these abstract concepts (maps or tables) are represented in computers using data structures. The models represented by data structures can be manipulated using reasoning algorithms to derive new information from known information. Therefore, the AI curriculum in primary and secondary schools should include content related to abstract thinking and data structures.
Figure 3 Scratch Game Design Stage for Pac-Man
(c) How to Learn from Data
Core machine learning algorithms in AI allow computers to create their own knowledge representations using training data provided by humans or acquired by machines themselves. If the context designed for operation and learning belongs to a limited domain, the performance of machine learning algorithms is quite accurate and reliable, enabling them to achieve very specific goals, such as AlphaGo’s victory in human-machine matches and IBM Watson’s effective treatment plans for tumors. These fields mainly involve mathematics, medicine, physical sciences, and computer science.
Machine learning is more suitable for applications based on rules, content, etc., such as facts, methods, operations, algorithms, and procedural skills, and is less likely to support complex, hard-to-evaluate skills such as critical thinking, effective communication, and causal explanation. The AI curriculum in primary and secondary schools should include basic machine learning algorithms, and students should be able to use open tools (such as RapidMiner, a visual programming platform for machine learning and AI based on drag-and-drop similar to Scratch) to perform basic machine learning operations such as classification, clustering, and recommendation.
For example, the Machine Learning for Kids[9] project introduces AI by combining IBM Cloud through APIs with Scratch, allowing students to train learning projects using IBM AI technology to recognize text, graphics, and speech, and integrate the trained results into Scratch’s online platform as extended modules, as shown in Figure 4. In the “Smart Classroom” case, a text model for “turning on the light,” “turning off the light,” “turning on the fan,” and “turning off the fan” was trained through machine learning and imported into Scratch, forming new programming “blocks,” as shown in the middle part of Figure 5, allowing the program to automatically turn on the light when the text “It’s too dark” is entered in Scratch, as shown on the right side of Figure 5 (the left side shows the state before command execution).
Figure 4 Training Text Models in Machine Learning for Kids
Figure 5 AI “Smart Classroom” Case
(d) How to Collaborate with AI
When collaborating with AI, students must understand and perform at least three key tasks: training AI applications to perform certain tasks; explaining the results of these tasks, especially when the results are counterintuitive or controversial; and using these AI applications responsibly.
First, machine learning algorithms must be taught how to complete their designed tasks. In this work, large training datasets are used to teach machines how to perform their intended tasks. For example, medical AI programs are used to detect diseases, and recommendation engines support learning material recommendations and learning path planning, all of which require accumulating large amounts of data in advance and “teaching” AI how to recognize and process this data.
Second, AI applications generally reach conclusions through opaque processes (the so-called black box problem), so students need to understand that human experts in the field should explain the behavior of AI to non-expert users and how they arrive at final results, such as experts needing to help insurance companies and law enforcement understand why autonomous vehicles take actions that lead to accidents.
Finally, it is essential to ensure that AI systems operate normally and safely. For example, AI systems should protect user privacy, ensure that the data input into AI systems complies with the General Data Protection Regulation (GDPR), and industrial robots working with humans should be designed to recognize nearby humans without endangering them.
(e) How to Evaluate AI Applications
Students should recognize that AI can both benefit and harm society. They need to view the application of AI in social life dialectically and understand what ethical standards AI systems should meet, as well as the ethical construction of AI systems that impact people’s lives, which needs to focus on issues of transparency and fairness. For example, AI systems should not discriminate against specific groups, and older students should be able to assess the potential ethical or social impact issues that new AI technologies may lead to.
5. Suggestions for Course Implementation
The implementation of AI courses should be flexible and diverse to accommodate students of different ages and learning abilities. Through experiences, practices, explorations, and innovations, students can better understand and apply AI, preparing them for the future.
(a) Experiencing Intelligent Life
Students’ growth is a process of gradually establishing fixed connections, constantly developing towards adapting to the environment. However, if tools are introduced too early, they may limit their imaginative expression, losing more possibilities. Therefore, for lower-grade elementary school students, tools should not become obstacles to their imagination. The focus of AI courses for lower-grade elementary students should not be on teaching them coding and creating their own AI programs, but rather on helping them understand what AI applications exist in daily life, how humans interact with AI, and what conveniences AI brings to our production and life. The goal is to familiarize students with AI as much as possible, cultivate their interest in AI, and enhance their willingness to learn.
(b) Understanding Intelligent Applications
For upper-grade elementary and junior high school students, who already possess a higher cognitive level, the focus should be on helping them deeply understand how AI works and encouraging them to complete creative exercises. Students should be encouraged to engage hands-on, addressing real-life problems or existing application prototypes, and to express their rich imaginations through graphical programming platforms.
For students at this stage, it is recommended to use comprehensive platforms similar to Blockly (a fully visual modular programming website). For example, the Machine Learning for Kids[9] project, supported by the MIT Media Lab’s Personal Robotics Group, and platforms like Cognimates[10], and Ecraft2learn[11] developed based on Snap and Snap4Arduino. These platforms encapsulate commonly used machine learning application module algorithms into a drag-and-drop component library, making AI application modules easily accessible. The encapsulated content is primarily based on six different areas of foundational knowledge—visual recognition, object manipulation, facial recognition, speech generation, speech recognition, and landmark-based navigation—equipping them with basic functions such as “seeing,” “hearing,” and “speaking,” along with interaction and navigation with the physical world, enabling basic intelligence.
(c) Participating in Intelligent Innovation
For high school students, typical applications of AI in personal assistants, autonomous driving, e-commerce, intelligent security, education, finance, and healthcare should be introduced, helping students understand how AI contributes to people’s lives. Additionally, the shortcomings of current AI applications and real-life problems encountered should be discussed and explored, encouraging active participation in teacher or related team projects to gain more foundational knowledge about AI, especially related to machine learning. For this purpose, the course practice platform can use a combination of Python and Raspberry Pi to complete the machine learning portion of basic AI learning. For example, a Python language platform like Google Colaboratory (abbreviated as Google Colab, a free research tool for the field of machine learning opened by Google) can be used to implement a small part of functionality in the project. By actively and dialectically evaluating AI applications, students can gain a more comprehensive understanding and application of AI knowledge.
References:
[1] AI automation could take over 50% of today’s work activity by 2045: McKinsey[EB/OL]. (2023-06-15)[2023-07-06]. https://cointelegraph.com/news/ai-automation-half-work-activities-by-2045-mckinsey.
[2] Report says AI could potentially replace 85 million jobs worldwide by 2025—are interns on the list?[EB/OL]. (2023-01-24)[2023-07-06]. https://news.yahoo.com/report-says-ai-could-potentially-203330619.html.
[3] Yang Du Jing, Huang Rong Huai, Li Zheng Xuan, et al. The Connotation and Construction Principles of AI Ethics in the Era of Intelligent Education [J]. Research on Educational Technology, 2019(6):1-9.
[4] Academician Pan Yunhe: Why do we say that AI has entered 2.0?[EB/OL].(2018-09-30)[2019-04-22].http://www.sohu.com/a/257300590_725934.
[5] SAXTON D, GREFENSTETTE E, HILL F, et al. Analysing mathematical reasoning abilities of neural models[C]//The International Conference on Learning Representations 2019. arXiv preprint arXiv, 2019:1-17.
[6] GILSTER P, GLISTER P. Digital literacy[M]. New York: Wiley Computer Pub, 1997.
[7] BAYKOUCHEVA, SVETLA. Managing scientific information and research data[M]. Cambridge: Chandos Publishing, 2015.
[8] AI4K12-sparking curiosity in AI[EB/OL]. (2022-12-14)[2023-07-07]. https://ai4k12.org/.
[9] Machine learning for kids[EB/OL]. (2023-06-29)[2023-07-07]. https://machinelearningforkids.co.uk.
[10] Home-Cognimates[EB/OL]. (2023-06-20)[2023-07-07]. http://cognimates.me/home/.
[11] A guide to AI extensions to snap![EB/OL]. (2022-03-20)[2023-07-07]. https://ecraft2learn.github.io/ai/.
Reflections on the Content of Artificial Intelligence Teaching in Primary and Secondary Schools in the Context of Core Literacy
Faqin SUN1, Nuo XU2
（1.School of Journalism and Communication, Yangzhou University, Yangzhou 225009, Jiangsu;
2.Suzhou University Experimental School, Suzhou 215133, Jiangsu）
Abstract: With the rise of the new round of scientific and technological revolution and industrial revolution, great changes have taken place in the demand for talents, and new requirements have been put forward for education. It has become a general trend to set up artificial intelligence related courses in primary and secondary schools. However, at the present stage, the confusion of curriculum content, the tradition of curriculum form, the singleness of curriculum output and the lack of faculty have adversely affected the popularization and promotion of artificial intelligence. The research focuses on the core literacy (scientific literacy, new digital literacy and new humanistic literacy) in the era of artificial intelligence. This paper introduces how the teaching content of artificial intelligence curriculum in primary and middle schools should be organized from five aspects: how to perceive the world, how to represent the world, how to study and apply knowledge and how to evaluate the application of artificial intelligence. It also expounds how the curriculum should be integrated with daily life from three levels: feeling intelligent life, understanding intelligent application and participating in intelligent innovation.Keywords: Artificial intelligence; Teaching content; Primary and secondary schools
Editor: Wang Tianpeng Proofreader: Wang Xiaoming
Previous Issue Directory｜Hyperlink Version
《China Educational Informatization》2023 No. 7
《China Educational Informatization》2023 No. 6
《China Educational Informatization》2023 No. 5
《China Educational Informatization》2023 No. 4
《China Educational Informatization》2023 No. 3
《China Educational Informatization》2023 No. 2
《China Educational Informatization》2023 No. 1
Click here for online submission
Journal Introduction
《China Educational Informatization》was founded in August 1995 and is a national academic journal supervised by the Ministry of Education of the People’s Republic of China and hosted by the Educational Management Information Center of the Ministry of Education, publicly issued at home and abroad. The journal aims to comprehensively and deeply record the progress of educational informatization construction, research and application results, and industry development status in China, conduct research on the development status of educational informatization in China and international comparative studies, and serve national macro decision-making in education; it strives to accurately grasp the policies and standards related to educational informatization, track the development trends of cutting-edge ICT technologies and their integration with education, and deeply explore the institutional and mechanism innovations in educational informatization construction and application, serving the practice of educational informatization in China.
● RCCSE Core Academic Journal of China
● Chinese Humanities and Social Sciences (AMI) Expanded Journal
● Journal included in the China Journal Full-text Database
● Journal included in the Chinese Science and Technology Journal Database
● Journal included in the National Philosophy and Social Science Academic Journal Database
● Journal included in the Core Journal (Selection) Database of China
● Journal included in Longyuan Journal Network
Published 12 issues a year, on the 20th of each month
20 yuan / issue, Postal code: 82-761
Editorial office email: [email protected]
Correspondence address: Room 921, No. 2 Huayuan North Street, Xicheng District, Beijing