In today’s world, the technological revolution and industrial transformation are deeply intertwined, and the development of an innovative nation has placed entirely new demands on the qualities of talent. The accelerating penetration of artificial intelligence has compelled the education sector to fundamentally reexamine the core question of “what kind of people to cultivate and how to cultivate them”. In an era where knowledge is increasingly accessible and standard answers can be generated instantly, science education can no longer focus on rote memorization and formulaic procedures; instead, it must return to the exploration of the unknown, the discovery of problems, and the reconstruction of failure. It is precisely against this backdrop that the Ministry of Education’s issuance of the Guidelines for the “Learning by Doing” Pioneering Initiative in Science Education during Compulsory Education (hereinafter referred to as the Guidelines) is most timely. Drawing on my long-term experience in chemical research and my practical involvement in science popularization and talent cultivation in primary and secondary schools, I would like to share a few insights regarding the spirit and implementation pathways of the Guidelines.

I. Identifying the Problem: When “Science Class” Is No Longer Like “Doing Science”

In recent years, the infrastructure for science education in primary and secondary schools has improved significantly, and students’ scores on paper-and-pencil tests have continued to rise. However, a visit to an actual science classroom reveals that the deep disconnect between science instruction and scientific research remains striking. Its typical symptoms can be summarized as “proceduralized operations, standardized conclusions, and prohibition of failure”. Students simply follow predetermined steps to complete experiments, record expected phenomena, and fill in standardized conclusions to successfully “pass”. The entire process rarely involves independent exploration of relationships between variables, seldom prompts students to question the significance of anomalous data, and does not allow for adjustments to experimental designs or iterative processes—yet it is precisely these elements that constitute the everyday practice of scientific research.

As is well known, the norm in scientific research is to approach the truth through trial and error and to seek opportunities in anomalies. Yet current science education has reduced this generative, reflective process of exploration to a formulaic approach that judges results as “right” or “wrong”. Students learn only the ability to reproduce predetermined answers, rather than the critical thinking and adaptability needed to navigate an uncertain world.

The advent of artificial intelligence has further amplified the urgency of this issue. The ingenuity of the Guidelines lies in their focus on the very core of the classroom: science education must shift from a “knowledge-transmission” model to a “practice-based construction” model, enabling students to truly engage in hands-on scientific activities rather than merely “recounting” scientific concepts in writing. This is not merely an adjustment to teaching methods, but a profound transformation in the logic of education.

II. Grade-level Framework: Aligning with Cognitive Development to Establish a Gradual Progression in Scientific Practice

The development of scientific thinking must be grounded in the process of solving problems through firsthand experience. The Guidelines clearly advocate for “student-centered learning and practice-based approaches”, which essentially means returning the initiative in inquiry to students and restoring science education to its proper logic of “deriving from practice and returning to practice”.

In terms of educational principles, the Guidelines demonstrate a distinct problem-oriented approach and a humanistic perspective. At the value level, they emphasize cultivating a scientific character—one that is rigorous, pragmatic, and unafraid of failure—through hands-on activities; at the student development level, they insist on protecting curiosity and respecting individual differences; and at the implementation level, they advocate breaking down the walls of the campus to effectively bridge classroom knowledge with real-world research contexts. These three principles mutually reinforce one another and collectively underscore that science education must respect facts, respect the process, and respect growth.

In terms of grade-level design, the Guidelines reflect a clear cognitive progression. The elementary school stage focuses primarily on concrete observation and simple hands-on activities, emphasizing the stimulation of interest and the accumulation of intuitive experience; the middle school stage progresses to independent design of experiments, control of variables, analysis of errors, and attempts at reflection and iteration, corresponding to the critical period for the development of abstract thinking and training in the scientific method. This arrangement is highly consistent with my personal experience in scientific research—that is, the development of scientific competence must follow a spiral progression from “observing” to “doing” and then to “thinking”.

III. Implementation Path: Systematic Restructuring of Class Hours, Teaching, Assessment, and Collaboration

A close reading of the Guidelines reveals a strong emphasis on practical implementation. It provides actionable institutional arrangements across four dimensions: class hour allocation, teaching transformation, assessment reform, and social collaboration.

Regarding the guarantee of class hours, the Guidelines explicitly require that students in grades 4 through 9 complete at least one full inquiry-based practical assignment each semester, lasting no fewer than four class hours, while no additional requirements are imposed on students in grades 1 through 3. This provision secures dedicated time and space for science-based practical activities at the institutional level, while the absence of mandatory requirements for lower grades reflects respect for children’s cognitive development.

In terms of teaching methods, the Guidelines center on six major themes—life and health, the ecological environment, the mysteries of the Earth, aerospace, emerging industries, and artificial intelligence—to guide schools in conducting interdisciplinary, project-based, and real-world science inquiry activities. Take the carbon dioxide production experiment as an example: traditional teaching often follows a step-by-step approach that presents conclusions first. Redesigned according to the “learning by doing” philosophy, students must independently experiment with different combinations of acid concentrations and carbonate quantities, observe changes in gas yield, identify leaks in the apparatus, and analyze the reasons for differences in the products. This is, in fact, the embryonic form of scientific research involving variable control, evidence-based reasoning, and iterative refinement of hypotheses. The teacher’s core task is no longer to provide a standard procedure, but to guide students in building their own cognitive frameworks through trial and error.

Regarding assessment reform, the Guidelines clearly state that no new paper-and-pencil exams for science special projects will be introduced; instead, performance during the inquiry process will be incorporated into comprehensive quality portfolios, with a focus on evaluating problem-solving awareness, hands-on practical skills, reflection and improvement, and teamwork. Even if the final experimental results are not ideal, students should receive positive recognition as long as they can reasonably explain the causes of deviations and propose ideas for improvement. This assessment approach—which “emphasizes thinking over results”—reflects respect for the inherently forgiving nature of science and serves as a profound response to the reality of the AI era, where “conclusions are easy to obtain, but critical thinking is hard to come by”.

In terms of collaborative education, the Guidelines encourage universities, research institutes, and high-tech enterprises to open their facilities to elementary and secondary schools for shared use. The cognitive experience gained from conducting an experiment firsthand far surpasses that of passively memorizing knowledge; when frontline researchers share a real-life account of a research failure, its educational impact often exceeds that of a carefully scripted “success story”. Breaking down the walls of the campus is not only an effective strategy for overcoming resource constraints but also a vital pathway for deep immersion in scientific culture.

IV. Responding to the Times: Cultivating Scientific Literacy That Brings Certainty Amid Uncertainty

Artificial intelligence is becoming deeply involved in knowledge production, and the value of “standard answers” is rapidly diminishing. What is needed to build an innovative nation is no longer talent that relies on memorization and replication, but rather multidisciplinary innovators capable of identifying problems in complex situations, distinguishing truth from falsehood amid a flood of information, and converging on solutions through trial and error and iterative refinement. The strategic positioning of science education in compulsory education must be elevated from “stockpiling knowledge for subsequent learning” to “shaping scientific ways of thinking and character traits for lifelong development”.

The profound value of the Guidelines lies in guiding a fundamental shift in education: moving science classes from “memorization and reproduction” to “inquiry and dialogue”, from “standardized responses” to “solving real-world problems”, and from “pursuing perfect results” to “respecting the factual process”, thereby truly bringing about a fundamental transformation in educational philosophy.

To implement the Guidelines, there is no need to deliberately create a flashy classroom environment; what is required is a solid, step-by-step approach starting with every experiment and every inquiry activity. We should encourage students to get hands-on, allow them to make mistakes, and guide them in reflection. Let “learning by doing” become the norm in science classrooms, and let science education return to its practical roots. Only in this way can we, in a future landscape shaped by the rapid evolution of artificial intelligence, nurture a generation of builders who possess scientific rationality, humanistic sensibilities, and the courage to innovate—thereby laying the most crucial talent foundation for building an innovative nation.