The Boundaries and Pathways of AI-Empowered Chemistry Education in K-12 Education

—A Systematic Restructuring of Classroom Instruction, Laboratory Instruction, Assessment System, and Teacher Education

Fang Yu

(Virtual Teaching and Research Office for Chemistry (Teacher Education) at Western Universities (Ministry of Education)   School of Chemistry and Chemical Engineering, Shaanxi Normal University   Xi'an, Shaanxi   710119)

Since the release of ChatGPT in late 2022, the impact of artificial intelligence (AI) on education has rapidly evolved from a “future prospect” to a “real-world challenge”. In just a few short years, a wide variety of AI-assisted teaching tools have emerged — ranging from personalized learning path planning to intelligent homework grading, and from virtual experiment simulations to knowledge graph construction — and the pace at which technology is penetrating K-12 education has far exceeded expectations. In late 2024, the Arizona State Board for Charter Schools approved the establishment of a new online charter school, Unbound Academy^[1]. Starting in the fall of 2025, AI will complete the daily academic curriculum for students in grades 4 through 8 within two hours, while the remaining four hours will be dedicated to teacher-led instruction, project-based learning, and workshops on public speaking and financial literacy — all designed to strengthen critical thinking and comprehensive skill development. Although this initiative is controversial, it signals an unavoidable trend: when AI can deliver knowledge with extreme efficiency, the structure of classrooms and the role of teachers in K-12 education will inevitably undergo a transformation.

In January 2025, the “Outline of the Plan for Building a Leading Country in Education (2024–2035),” issued by the Central Committee of the Communist Party of China and the State Council, explicitly called for “promoting the use of artificial intelligence to drive educational transformation” ^[2]. Due to its unique nature, the discipline of chemistry faces particularly significant challenges in this transformation. Chemistry education must not only help students establish a microscopic cognitive framework ranging from atoms and molecules to chemical bonds but also cultivate a scientific spirit and rational attitude through the discipline’s distinctive modes of thinking. Currently, AI is still unable to replace the in-depth teaching required to cultivate a scientific spirit and a rational attitude; however, if teachers do not proactively adapt, they risk being marginalized in a technological wave that prioritizes efficiency above all else. This article will explore three dimensions — the challenges AI poses to basic chemistry education, the deep-seated issues that must be addressed, and teacher training and professional development — and offer recommendations based on this analysis.

1 Challenges Posed by AI to Basic Chemistry Education

1.1 The Efficiency and Limitations of AI-Enhanced Learning

At the end of 2023, to gain a systematic understanding of millimeter-wave communication and terahertz detection technologies, the author conducted more than three hours of intensive study using ChatGPT-4 ^[3]. After more than thirty rounds of iterative questioning, and at the author’s request, it organized the entire interaction into two clearly structured and substantive PowerPoint presentations, categorized under the sections “Millimeter-Wave Communication” and “Terahertz Technology”. The efficiency and precision of the entire process far exceeded expectations. However, I also discovered that if questions were not precise or in-depth enough, or if the iterative process was insufficient, the AI was prone to providing vague or even inaccurate answers. This experience has convinced me that while AI-enhanced learning is highly efficient, “learning how to learn” has become a prerequisite. If students lack the ability to ask precise questions and engage in critical analysis, AI will not only fail to provide effective support but may actually lead them further down the path of misconceptions.

1.2 The Boundaries and Pathways for Bringing AI into the Classroom

AI can enable personalized instruction, dynamic visualization, and precise assessment of student progress — all of which are difficult for traditional classrooms to achieve. However, increased efficiency does not necessarily equate to improved educational quality. When a middle school student “learns” the law of conservation of mass in three minutes with the help of AI, memorizes the statement, and answers multiple-choice questions correctly, does he truly understand it? The teaching of chemical concepts has its own unique characteristics. The discovery of the law of conservation of mass is a classic example of how Lavoisier refuted the phlogiston theory^[4]; the scientific spirit and methodology embedded in this process are far more significant than the literal wording of the law itself. If AI-driven instruction focuses solely on “test points” while neglecting the “story” behind them, and merely outputs conclusions while omitting the process of inquiry, students will gain nothing more than the empty shell of knowledge — rather than the lifeblood of scientific thought.

A practical case study involving a middle school chemistry teacher that the author encountered provides valuable empirical evidence for AI-assisted classroom instruction. In the “Law of Conservation of Mass” unit, the teacher used AI to support students’ independent learning of basic concepts before class. Classroom time was no longer devoted to conventional lecturing but was instead focused on a core cognitive conflict — namely, an in-depth discussion of the question, “Why does the mass of a magnesium rod increase after combustion?” The teaching outcomes showed that students significantly outperformed those in traditional lecture-based classes in both the depth of their conceptual understanding and their ability to apply concepts to new problems. The key insight from this case is that the teacher converted all the classroom time freed up by AI into inquiry-based dialogue and critical thinking training, rather than guiding students to mechanically rewatch AI explanations and simply add more practice exercises. The latter approach essentially merely replaces the medium through which information is conveyed with a “cold medium” that lacks pedagogical interaction and emotional support. It is therefore evident that the role of AI in chemistry classrooms should be that of a learning tool that facilitates the expansion of students’ cognitive horizons, rather than a digital lecture platform that compresses the thinking process.

In the era of AI, the importance of chemistry laboratory instruction should not be diminished but rather strengthened. Laboratory instruction serves a function that theoretical instruction cannot replace: students gain sensory experiences — such as changes in color, temperature, gas production, and precipitate formation — through hands-on activities, which no virtual simulation can fully replicate. The uncertainties inherent in the experimental process — such as anomalous results caused by operational errors or new questions raised by unexpected phenomena — can actually serve as opportunities to cultivate students’ abilities to observe carefully, analyze deeply, and adapt flexibly. Hands-on experiments also allow students to experience the entire process of scientific inquiry firsthand, thereby helping them understand the practical origins of scientific knowledge — something that virtual scenarios struggle to replicate. Of course, AI can play a supporting role in laboratory instruction: virtual simulations can help students preview procedural steps and understand experimental principles; utilize intelligent analysis to assist in interpreting experimental data and diagnosing operational issues; and employ safety alerts to prevent hazardous procedures. However, AI cannot replace real-world experimental procedures. Furthermore, chemistry experiments often require group collaboration, a process that helps cultivate students’ sense of cooperation, responsibility, and adherence to standards — educational values that AI cannot provide. Therefore, the correct approach to AI-enabled laboratory instruction is to use AI to augment the effectiveness of experimental teaching, rather than replacing the real experience with a virtual one.

1.3 Essential Skills for the AI Era: Learning to Learn

“Learning to learn” is the most essential skill in the AI era, yet it has not been given sufficient attention in basic education. In an era of information scarcity, the core mission of education was to help students “acquire knowledge”; in the AI era, with knowledge acquisition more convenient than ever, the core mission of education must shift to helping students “learn how to learn”. Based on the author’s experience using ChatGPT-4, “learning to learn” encompasses at least four dimensions: First, clarifying one’s own needs and accurately defining the boundaries of knowledge; Second, distilling those needs into precise questions; the more precise the questions and the deeper the iteration, the more valuable the AI’s responses will be; third, maintaining a critical perspective on AI outputs, as users must independently assess their scientific validity and credibility — and in chemistry, as an empirical science, experimental verification remains an irreplaceable means of validation; fourth, effectively using AI to structure fragmented knowledge and integrate it into a systematic cognitive framework.

My experience in graduate education has also confirmed this. Since 2013, I have required all incoming master’s and doctoral students to take a MOOC course on sensors or molecular materials offered by a prestigious international university and pass the exam. For more than a decade, students in every cohort have found the course “particularly challenging”, but they have all persevered in the end. They have gained not only an enhancement of their professional knowledge and English proficiency but also a broadening of their horizons and a boost in self-confidence. This has led me to deeply realize that guiding students to a state where they can independently acquire knowledge is far more important than simply imparting specific knowledge to them. This principle applies equally to basic education. However, the reality is that students at the basic education level generally lack training and habits in self-directed learning, yet “learning how to learn” is precisely the core competency most urgently needed in the AI era — one that must be cultivated from an early age. The growth journey of mathematician Wang Hong, one of this year’s Fields Medal recipients, also corroborates this point. According to reports, she developed the habit of independent learning from an early age; this ability has permeated her academic career and served as a crucial foundation for her outstanding achievements ^[5].

2 Three Deep-Seated Issues We Must Confront

2.1 Redefining the Role of Teachers

The role of teachers must be redefined, but “transformation” does not mean “withdrawal”. When the transmission of knowledge can largely be entrusted to AI, the unique value of teachers lies in providing moral guidance, emotional support, and intellectual stimulation. Specifically in chemistry education, after AI has covered foundational topics such as elements, atoms, molecules, atomic structure, chemical bonds, and reaction rates, an excellent chemistry teacher should guide students to think deeply beyond basic concepts: How was Mendeleev able to predict unknown elements? What is the mindset behind scientific discovery? From semiconductor materials, new energy batteries, biopharmaceuticals, and specialty functional materials, to green synthesis and catalytic conversion under the “dual carbon” goals — how does chemistry underpin these critical fields that concern national strategy, the sustainable development of human civilization, and the well-being of the people? What is the relationship between scientific research and social responsibility? Such teaching transcends the mere transmission of knowledge and enters the realm of critical thinking and value guidance.

Recently, scholars Li Xijuan, Zhu Yujun, and Wang Lei co-authored a paper ^[6] pointing out that new technologies, represented by AI, are profoundly transforming scientific research paradigms, social operating models, and forms of knowledge production, thereby necessitating a systematic restructuring of educational goals, teaching models, and disciplinary development. In this transformation, basic education is on the front lines, as this stage is precisely the critical period during which students’ worldviews, outlooks on life, and values are formed. Chemistry education embodies unique educational value: it not only imparts knowledge about material transformations but also, through the discipline’s distinctive modes of thinking, cultivates students’ scientific spirit, rational attitudes, and commitment to truth-seeking. These qualities are precisely what AI cannot easily replace, and they represent the core ground that teachers must defend in the AI era. However, it must also be recognized that if teachers do not proactively transform their practices — stepping away from the rote delivery of knowledge to focus on educational activities of higher value — then the replacement of teachers by AI will no longer be merely a hypothetical scenario at the technical level, but simply a matter of time.

2.2 Systematic Restructuring of the Evaluation System

The failure of the evaluation system is the most profound challenge posed by AI. If students can easily complete homework and tests with the help of AI, can exam scores still reflect genuine learning outcomes? If acquiring knowledge becomes so convenient, how meaningful are assessments that focus on memorization and reproduction? The question is not “whether to ban students from using AI”, but rather “what is truly worth evaluating”: Is it how many chemical reaction equations they have memorized, or their ability to design experimental protocols to solve real-world problems? Is it the ability to recite definitions of concepts, or the ability to apply chemical principles to analyze social issues? This requires shifting the focus of assessment from “testing knowledge retention” to “assessing competencies and literacy”, from “summative assessment” to “formative assessment”, and from “single-dimensional” to “multidimensional indicators”. Some teachers have already experimented with a “human-machine collaboration” assessment model: students are allowed to use AI tools when completing open-ended tasks, but they must clearly indicate the AI’s contributions and their own independent thinking. The assessment emphasizes problem analysis, solution design, and critical reflection.

Teacher evaluation also faces challenges. If knowledge transmission is primarily handled by AI, and teachers are then evaluated based on students’ test scores, this is not only unfair but also lacks a theoretical foundation. We need to explore a new teacher evaluation system that incorporates curriculum design skills, learning facilitation skills, and student development guidance skills into the evaluation framework. School administration also requires a systematic overhaul: Is a standardized curriculum schedule still necessary? Is the traditional classroom-based instruction model still the only organizational form? As personalized learning becomes the norm, a series of supporting systems — including credit recognition, examination systems, and the calculation of teacher workloads — will need to be redesigned. Textbook development is also undergoing a transformation: Printed textbooks are no longer the primary vehicle for knowledge; future chemistry textbooks may increasingly serve as learning roadmaps and problem sets.

2.3 Upholding the Bottom Line in AI-Empowered Education

In the process of advancing AI-enabled education, there are several bottom lines that must be upheld. First, we must guard against the risk of mental inertia. If students become accustomed to obtaining ready-made answers from AI, their ability to think independently may deteriorate. The appeal of chemistry lies in exploring the unknown; if students merely passively accept conclusions and no longer ask “why” or “are there other possibilities,” the very essence of chemistry education will be lost. Second, we must be vigilant against the widening of the digital divide. Access to AI educational resources depends on hardware and internet connectivity; if high-quality resources are concentrated in developed areas and top-tier schools, the educational gap between urban and rural areas — as well as across regions — may widen further. Third, we must not stray from the essence of education. Efficiency is not the most important goal of education. Instilling values, shaping character, and fostering emotional growth require us to slow down and engage in deep, person-to-person interaction. Students in basic education are in a critical phase of character development; they need not only highly efficient knowledge input but, more importantly, a nurturing environment for growth. Fourth, we must guard against oversimplifying the role of teachers. A teacher’s core value lies precisely in what AI cannot do: observing students’ emotional changes, seizing unexpected moments in the classroom, discovering students’ unique qualities, and helping them build self-confidence. These educational practices — which are difficult to quantify — are the true embodiment of a teacher’s professionalism.

3. Teacher Training and Professional Development Empowered by AI

Discussions on AI-enabled basic chemistry education have largely focused on the instructional level, but two more fundamental issues have been widely overlooked: the lag in the training of pre-service teachers and the challenges faced by in-service teachers in continuing their professional development. In a sense, these two issues deserve more attention than the AI tools themselves, as they will determine the quality of chemistry education at the basic education level for quite some time to come.

3.1 The Lag in Teacher Education and Its Restructuring

The lag in teacher education curricula represents a latent crisis. Currently, chemistry programs at teacher training institutions generally still follow the basic framework of “subject knowledge instruction + study of educational theory + teaching skills training + teaching practicum.” This framework was reasonable in the past, as teacher candidates needed to acquire sufficient knowledge of chemistry before stepping into the classroom. However, in an era where AI can efficiently impart knowledge, should the focus of teacher education be adjusted?

The problem lies not only in the content and teaching methods of specialized courses, but also in the severe lag in pedagogy courses. Current educational theory studies and teaching skills training still assume a traditional classroom setting, and core courses — such as instructional design, classroom organization, and student assessment — hardly take AI into account. Prospective teachers learn how to design instructional processes, how to ask questions, and how to organize discussions, but they are rarely guided to consider: When AI can instantly answer students’ questions, how should teachers adjust their questioning strategies? When every student has a personalized AI learning companion, how will the foundation and methods of classroom discussions change? When AI can automatically grade assignments, how should the function and form of assessment be redefined? These are not issues of subject-specific knowledge, but rather issues that require a fundamental restructuring of pedagogy itself.

The foundation of subject knowledge also needs to be reevaluated. Pre-service teachers still need to possess a solid foundation in chemistry, but the meaning of “solid” is changing. In the past, “solid” meant memorizing a large number of facts, formulas, and reactions and being able to recall them at any time; in the future, “solid” will increasingly mean a deep understanding of the discipline’s core concepts and methodological approaches, the ability to assess the accuracy of AI-generated content, and the ability to translate abstract principles into problem scenarios that students can relate to. The specific details of knowledge that AI can provide at any time do not necessarily need to occupy the teacher candidates’ valuable memory capacity; conversely, the subject-specific insight and judgment that AI cannot replace must be reinforced even more.

Reforming the content of pedagogy courses is more urgent than adjusting specialized coursework. Chemistry pedagogy courses require a systematic update: How can student learning be assessed in an AI-assisted environment? How can AI be used to support differentiated instruction and personalized guidance? How can inquiry-based learning tasks involving human-AI collaboration be designed? How can students be guided to critically evaluate answers provided by AI? These topics should form the core modules of teaching competency training for the next generation of pre-service teachers. If pedagogy courses themselves do not undergo innovation, pre-service teachers will enter new classrooms equipped with an outdated toolkit—and this is precisely the crisis that warrants the greatest vigilance in the coming years.

Adjustments to teacher education cannot be limited to adding one or two elective courses on “AI tool usage”; instead, a more systematic approach is needed. First, prospective teachers should be given the opportunity to deeply experience the process of AI-assisted learning as learners themselves — just as the author did when using ChatGPT-4 to study millimeter-wave communication and terahertz detection. Only by personally experiencing the iterative process of “asking-answering-probing” and understanding the strengths and limitations of AI can they effectively guide students in their future teaching. Second, it is necessary to cultivate pre-service teachers’ instructional design skills in AI-assisted environments: when to have students ask questions directly to AI, when to organize face-to-face discussions, and when the teacher needs to provide explanations in person — these judgments require experience gained through practice. Third, pre-service teachers must be trained in classroom management skills within AI-assisted environments: When each student is at a different stage of AI-guided learning, how can teachers coordinate the instructional pace for the entire class? How can they ensure that classroom discussions remain effective? How can they monitor the learning progress of every student? These issues are rarely addressed in traditional pedagogy courses.

More fundamentally, teacher education needs to rethink the core competencies of chemistry teachers. In the future, an excellent chemistry teacher may no longer need to memorize all chemical equations (since AI can provide them at any time), but must be able to design learning scenarios that inspire students’ curiosity; must be able to identify the confusion students encounter during AI-assisted learning and provide effective guidance; and must be able to help students develop critical thinking and scientific judgment after they have been exposed to a large amount of AI-generated information. Cultivating these competencies requires systematic adjustments to the teacher education curriculum, rather than piecemeal fixes. The inertia of teacher education is formidable, and adjusting training programs takes time — but the rapid spread of AI will not wait.

3.2 Challenges and Solutions for the Continuing Education of In-Service Teachers

The challenges facing in-service teachers’ continuing education represent an immediate crisis. Most of the hundreds of thousands of chemistry teachers currently working in K-12 education were trained under a traditional educational model centered on the teacher and focused on the transmission of knowledge. Although many of them have heard of AI, they lack systematic training and practical experience. A common phenomenon is this: schools introduce AI teaching platforms, and teachers attend training to master basic operations; yet, once back in the classroom, they continue to use their original teaching methods, treating AI merely as an “auxiliary tool” used only occasionally — its potential remains far from being fully tapped. This situation does not stem from a lack of effort on the part of teachers; rather, the transformation brought about by AI-empowered education involves every aspect of teaching, and training in operational skills alone is far from sufficient. Teachers need to redefine their roles, redesign curriculum structures, and rethink assessment methods — all of which cannot be addressed through superficial training alone. Even more concerning is that, when promoting AI-based instruction, some schools simply require teachers to “just use it”, without establishing systematic institutional follow-up, providing ongoing academic support, or forming the necessary learning communities.

Currently, most AI training programs for teachers are limited to “feature introductions and operational demonstrations”, which are disconnected from real-world teaching scenarios. Teachers need more than a general understanding; they require specific guidance tailored to the unique characteristics of chemistry as a subject. For example, how can AI be used to explain atomic structure? How can AI be used to design tiered assignments? How can AI support laboratory instruction? How can teachers identify incorrect information provided by AI? These are the practical skills teachers truly need to master.

Establishing teacher learning communities is a viable approach. When teachers within the same school or district form learning communities and regularly share their attempts, challenges, and experiences with AI-assisted teaching, individual explorations coalesce into collective wisdom. One teacher may discover that using AI to organize discussions on the theme of “Chemistry and Daily Life” yields excellent results, while another may develop a method for designing tiered assignments using AI. Through communication within the community, these scattered experiences are consolidated, validated, and disseminated — a process far more dynamic than top-down, standardized training.

4. Recommendations and Outlook

Recommendation 1: Restructure teacher education programs to incorporate AI literacy into core curricula. Chemistry teacher education programs should maintain a solid disciplinary foundation while adding modules such as AI-assisted instructional design and human-machine collaborative classroom organization. This will enable student teachers to deeply experience AI-assisted learning as learners themselves, helping them understand the advantages, limitations, and impact of AI on learning methods before they step into the classroom. It is recommended that the Ministry of Education add requirements related to AI literacy to its accreditation standards for teacher education programs to drive curriculum adjustments at teacher training institutions nationwide. A specialized research project on “Core Competencies for Chemistry Teachers in the AI Era” should be launched to provide theoretical support for adjustments to training programs. At the same time, teacher training institutions should take the lead in developing model courses and case repositories for AI-assisted instruction, providing teacher candidates with practical resources they can observe and emulate.

Recommendation 2: Establish a support system for the continuous professional development of in-service teachers. Education administrative departments should incorporate AI literacy into the mandatory curriculum for teacher training, placing particular emphasis on targeted training for subjects such as chemistry, rather than providing general technical introductions. Schools and educational research departments should promote a transformation in teaching and research methods, shifting from the traditional “class observation and evaluation” approach to a “research-practice-reflection” model, and establish teacher learning communities to pool individual explorations into collective wisdom. Educational research activities and teaching competitions at all levels should promptly incorporate AI-assisted teaching capabilities into their evaluation criteria to serve as a guiding force. In particular, two extremes must be avoided: first, “using without researching” — where platforms are installed and equipment is provided but ongoing professional support is lacking; and second, “being afraid to use it or not knowing how to use it” — where educators feel at a loss in the face of the technological wave. It is recommended that special funding be allocated to support teachers in conducting action research on AI-assisted teaching, enabling them to grow through practical experience.

Recommendation 3: Systematically advance reforms to the evaluation system. Shift student assessment from testing knowledge retention to evaluating competencies and literacy, and incorporate formative and performance-based assessments into the formal framework. Explore a “human-machine collaboration” assessment model that allows students to use AI in open-ended tasks — provided they attribute the AI’s contributions — with the focus of evaluation placed on problem analysis, solution design, and critical reflection. Reform teacher evaluation by incorporating course design capabilities, learning facilitation skills, and student development guidance abilities into the system. At the school administrative level, curriculum and teaching arrangements should be restructured to reduce the weight of standardized tests and increase the weight of open-ended tasks and practical assessments. It is recommended that education authorities organize experts to develop evaluation guidelines for core competencies in chemistry in the AI era, providing a practical reference framework for frontline teachers and schools.

The foundational, pervasive, and supportive nature of chemistry means that the quality of chemistry education is directly linked to the cultivation of a nation’s scientific and technological competitiveness and to the sustainable development of human civilization ^[7]. Seizing the historic opportunity presented by AI-empowered education is of great significance for cultivating a new generation of young people who possess a sense of national pride, a sense of responsibility, and innovative capabilities. However, technology is ultimately just a tool; the essence of education lies in nurturing people. While AI can help students acquire knowledge more quickly, it cannot replace teachers in sowing the seeds of the scientific spirit in students’ hearts, igniting the spark of curiosity, or guiding their pursuit of values. Let us safeguard the true nature of education amid technological empowerment and anchor our educational direction amid transformation and innovation, so that we may cultivate a next generation capable of mastering AI tools while possessing independent character and a scientific spirit.

Without chemistry, there is no future; basic chemistry education has a bright future in the age of AI.

References

[1] MacKenzie Price. “AI System Will Drive Academics at New Virtual Charter School” [EB/OL]. (February 27, 2025) [accessed August 7, 2026]. https://2hourlearning.com/ai-system-will-drive-academics-at-new-virtual-charter-school-2/.

[2] Central Committee of the Communist Party of China and China State Council. The Central Committee of the Communist Party of China and the State Council Issue the “Outline of the Plan for Building a Leading Country in Education (2024–2035)” [N]. People’s Daily, January 20, 2025 (1).

[3] Fang Yu. Artificial Intelligence Empowering Education: Starting from an Independent Learning Experience Guided by ChatGPT-4 [J]. University Chemistry, 2025, 40(9): 1–4.

[4] Zhang Desheng, Xu Wanghua. A Concise Course in the History of Chemistry [M]. 2nd ed. Hefei: University of Science and Technology of China Press, 2017: 35–51.

[5] China Science Daily. Exclusive Interview with Wang Hong: From an “Unnoticed Student” in Peking University’s Mathematics Department to a Fields Medalist [EB/OL]. [2026-08-07]. https://www.peopleapp.com/column/3005274302450000761-4179.

[6] Li Xijuan, Zhu Yujun, Wang Lei. Practical Pathways for AI-Empowered Talent Development in Chemical Innovation and Transformation Directions for Chemistry Education Research [J]. Chemical Education (Chinese and English), 2026, 47(3): 128–129.

[7] Fang Yu. Chemistry: A Key Force Supporting the Sustainable Development of Human Civilization [J]. University Chemistry, 2025, (12): 72–75.

Citation:

Fang Yu. The Boundaries and Pathways of Artificial Intelligence in Chemistry Education at the K-12 Level: A Systematic Restructuring of Classroom Instruction, Laboratory Instruction, Assessment Systems, and Teacher Education [J]. Chemistry Education (Chinese and English), 2026, 47(19): 14-18.