Overall, this is a pioneering work of real significance. Across 266 pages, the book systematically builds a theoretical framework for AI general education, puts forward cognitive autonomy as its core concept, and forms a complete analytical system around four capacities — induction, generalization, judgment, and self-awareness — and six dimensions: source vigilance, evidence sensitivity, boundary awareness, trust calibration, cognitive agency, and responsibility judgment. The curriculum framework of "five content areas horizontally, three progressive stages vertically", together with the textbook contents for 44 primary-school lessons and 40 lower-secondary lessons and the practice handbooks in the appendices, shows a complete design running from theory to curriculum. I learned a great deal from studying it.
From the perspective of frontline teaching research in basic education, there is still room to improve between the book's theoretical framework and classroom practice. The main point is that, beyond making the theory coherent, we would like to see cases showing how it actually lands in practice. From a university standpoint, perhaps the more abstract, the more sophisticated, and anything too concrete looks low-level; but from the frontline standpoint, teachers' theoretical grounding needs strengthening and the path from theory to practice is long. Teachers want to know not only why they are doing something and what it should achieve — even more, they want to know what exactly they should do, and what kind of instructional design will reach the goal. So a concrete case that carries the theory all the way into practice is what teachers need, especially one mapped back to the theory, giving them a template to imitate first and innovate on later.
1. Review of and Suggestions on the Book's Overall Logic
1.1 The Breakthrough Contribution of the Theoretical Framework
The theoretical architecture is the book's most outstanding contribution. The authors raise the goal of AI general education from "knowledge popularization" to cognitive autonomy, and connect it historically to the general-education tradition set out in Harvard's 1945 "Redbook". This positioning is accurate and far-sighted: education in the AI era must not only teach students to "use" AI, but also develop their ability to keep their judgment, calibrate their trust, and take responsibility in complex cognitive environments. Particularly commendable is the book's critique of nine tendencies in current AI education practice (instrumentalization, technicalization, activity-orientation, competition-orientation, frontier-chasing, fragmentation, ethics-as-an-add-on, superficial assessment, and superficial teacher training), all of which are highly relevant to reality.
1.2 The "Middle Layer" from Theory to Practice Needs Strengthening
The book's logic runs: historical background (Part I) → theoretical foundations (Part II) → content framework (Part III) → stage-by-stage implementation (Part IV). This structure is academically complete, but between the "content framework" and "stage-by-stage implementation" a key "middle layer" is missing — namely, how to turn the abstract content framework into concrete classroom instructional design.
The book aims to support education from primary school through university, using the development and delivery of AI general education courses to help students build the core competencies of the AI era, with cognitive autonomy at the center. But how to put the theory into teaching practice, and how to help teachers — frontline teachers above all — actually "land" the theory, is a very important question. What basic-education teachers need most is not a framework of "what to teach" but demonstrations of "how exactly to teach".
For example, Part III proposes analytical frameworks such as "seven-step application analysis", "six-step cross-disciplinary integration", and "seven-step deliberation". These frameworks are theoretically strong, but the book does not provide a single complete classroom teaching case showing how they land in the classroom. I suggest adding some concrete cases to close the loop from theory to practice; this could also be done through "electronic resources" and the like, linking the theory to its implementation and giving frontline teachers a model to move from imitation to innovation.
Similarly, for developing the four capacities of induction, generalization, judgment, and self-awareness — how to develop them, and through what kinds of teaching activities and processes — there needs to be a practical model, which could be illustrated through concrete cases.
I therefore suggest adding a dedicated section at the start of Chapter 14 demonstrating how to turn the "seven-step application analysis" framework into the instructional design for a primary or lower-secondary lesson. This could include: taking "recommender systems" as an example, showing concrete teaching-activity designs for the same topic at the primary, lower-secondary, and upper-secondary stages, covering scenario setting, core questions, student activities, assessment tasks, and the expected cognitive-autonomy goals — above all, how to design the whole around cognitive autonomy. Such a "framework → classroom" demonstration would greatly increase the book's value as guidance for frontline teachers.
2. Analysis of the Guidance for AI General Education Courses in Basic Education
Analysis of Chapter 14, "Integrated Curriculum Design"
Chapter 14 is the chapter most closely tied to basic education. It states clearly that "progression across stages means deeper explanation, wider transfer, and higher demands for evidence and responsibility judgment; it cannot be reduced to ever-increasing algorithmic difficulty", and stresses that "cultivating interest does not mean avoiding important content". These two points offer important guidance for correcting two tendencies in current basic-education AI teaching: turning it into algorithm competitions, and "primary pupils only play and don't learn". We have set up a large-discipline department, one of whose key tasks is integrated, through-line education, aiming at early identification of talent and long-chain cultivation. Studying this chapter made the through-line logic of AI education much clearer to me.
It should be noted that, judging from how AI general education courses are currently being run, AI education at the basic-education stage relies mainly on the information technology curriculum, and information technology teachers are the main people delivering it. The 2022 compulsory-education curriculum standards have already established four core competencies: "information literacy", "computational thinking", "digital innovation and intelligence", and "information social responsibility". In terms of the zone of proximal development, what are the correspondences between the book's cognitive autonomy framework and these four competencies, where does it go beyond them, and where do they need to be integrated? This is one of the questions frontline teachers and teaching researchers care about most. I suggest discussing it specifically in that chapter or in the Overview — that is, the concrete alignment between the book's curriculum framework and the current Compulsory Education Information Technology Curriculum Standards (2022 Edition).
At the same time, according to the Ministry of Education's 2025 guidelines, teaching AI general education is not only the job of information technology teachers; it also requires cross-disciplinary integration and cooperation with universities, research institutions, and other organizations. Chaoyang District, for example, has used models such as standalone courses, integrated courses, and jointly delivered courses. In fact, schools are all still feeling their way. Ideally, the book's cases would answer these questions: given the limited class hours for AI education, how can the "five content areas" be fully covered within them? Which content can be integrated into other subjects (such as science, morality and law, or integrated practice)? Which must be completed in information technology classes? I suggest building on Table 14.2, "Progression of the Five Content Areas across Stages", by adding "suggested proportion of class hours" or "suggested implementation pathways" to make it more actionable.
In addition, looking at the textbook contents and so on, my impression is that the background knowledge required of teachers is quite high, and the cognitive level expected of students for their age is overestimated (Haidian, and especially the areas in and around the universities, have exceptionally rich educational resources and should not be taken as the benchmark). So some of the textbook content — for example, the historical narrative of the development of science for primary pupils, and the terminology related to the deep learning era — may be rather abstract for primary pupils.
Some Suggestions on Assessment
The book's discussion of assessment is scattered across chapters and is mostly stated as principles. At the basic-education stage, assessment is the weakest link. I suggest adding, at the end of the book or in an appendix, a "Guide to Developing Assessment Tools for Cognitive Autonomy", offering ready-to-use examples of assessment tasks, rubrics, and scoring rules. Assessment at the primary stage in particular should rely mainly on observation records, oral expression, and analysis of student work, rather than written exams or quantitative tests. Taking Chapter 14 as an example:
Chapter 14 sets the goal for the primary stage as "cultivating interest and scientific spirit", and explains that scientific spirit includes "respecting facts and evidence, being willing to verify, and understanding that scientific results come from accumulation, collaboration, failure, and correction". This positioning is exactly right. Recently the state has also placed great emphasis on science and technology education (it seems it used to be called science education), and this matches my own experience in developing AI courses: both information technology and AI are important subjects within science and technology education. But the chapter does not give enough guidance on how "scientific spirit" should be concretely realized at the primary stage. I suggest adding descriptions of concrete behaviors showing "scientific spirit" at the primary stage, for example: after trying face recognition, students spontaneously test "Can it still recognize me with a mask on?"; on seeing an AI-generated image, they ask "Is this real?"; after using a voice assistant, they discuss "How does it understand what I say?". Such concrete behavioral descriptions can help teachers understand what "scientific spirit" means in the classroom.
Table 14.3, "Progression of the Four Cognitive Capacities across Stages", is one of the most instructive tables in the book. But its descriptions remain at a fairly abstract level. For example, the primary-stage description of "self-awareness" is "noticing whether one is being credulous, leaking information, or treating the machine as a person" — how can this be observed and assessed in class? I suggest adding "observable student behaviors" for each capacity at each stage, i.e., linking behaviors to assessment rubrics. For example, observable signs of self-awareness at the primary stage could include "asking parents for permission before using a voice assistant". Such concrete behavioral indicators will help teachers turn abstract capacity goals into observable assessment tasks.
Below are some suggestions on intelligent tools:
Chapter 12, "Cross-Disciplinary Integration": Add Cross-Disciplinary Cases Usable in Basic Education
Chapter 12 proposes the "six-step cross-disciplinary integration" framework and illustrates it with fields such as mathematics, engineering, physical science, biology and medicine, earth and space, and society, culture and the arts. But these cases lean toward university and research settings and are of limited use at the basic-education stage. I suggest adding cross-disciplinary cases usable in basic education, for example: how AI can help identify plants in biology class (science + AI), how to analyze patterns of word use in classical poetry (Chinese + AI), and how to use data analysis to improve school sports training (PE + AI). Cases close to students' lives like these would better help basic-education teachers and students understand cross-disciplinary integration.
Chapter 13, "Risk, Ethics and Responsibility": Add a Child's Perspective
Chapter 13's analysis of AI risks is comprehensive, covering information forgery, privacy leaks, data bias, algorithmic fairness, automated decision-making, and more. But the chapter is written mainly from an adult and social-governance perspective, and does not discuss enough the concrete risk scenarios children and adolescents face in daily life. For example: data collection when primary pupils use learning apps, the way recommendation algorithms on social platforms shape their content preferences, and deepfake content they encounter in online games. Risk scenarios that directly concern children like these should be discussed more fully in that chapter or in Chapter 14.
Artificial Intelligence General Education: Cognitive Autonomy in an Intelligent Society is a pioneering work of real significance, providing a systematic theoretical foundation and curriculum framework for AI general education. From the perspective of teaching research in basic education, the book's core value lies in: making clear that AI education is not training in tool operation but general education concerned with developing the capacity for cognitive autonomy; setting out a concrete path for spiral curriculum design; and providing, in its rich appendices, textbook contents for reference.
Suggestions: first, add demonstrations of the transition from theoretical framework to classroom teaching; second, clarify the alignment with current curriculum standards; third, provide observable indicators of capacity and assessment tools.
I look forward to the book becoming an important reference for AI general education in basic education once it is formally published, and to more exchange and cooperation with the author team in future work.