Pupils use AI to create a multisensory vegetable exhibition

Caibao’s Adventures: multisensory, AI-supported project learning in the early primary grades  

Gao Yating, Zheng Yandong, Hu Hanlin, Zhu Jiayu, Deng Yao, Guo Xuewen  

Shenzhen Education Science Academy Experimental School (Guangming)  

Abstract:Shenzhen Education Science Academy Experimental Primary School (Guangming) has introduced the “Vegetable Adventure” AI-enabled project-based learning approach to address the issue of weak understanding of vegetables among younger students. Using various AI tools, the students independently planned and carried out a “Smart Interactive Vegetable Exhibition” for kindergarten children. By analyzing vegetables, creating IP characters, theme songs, and games, they created a multi-sensory learning environment. This project successfully implements the integrated model of “AI + Art + Nature Education”, effectively enhancing the students’ overall abilities. It provides a practical example for the digital transformation of early childhood education and the integration of the five aspects of education.    

Keywords:AI-driven education; cross-age curatorial practices; multisensory integrated learning    

In the practice of vegetable cultivation on campus, schools have found that young children generally have a vague understanding of vegetables and little interest in them. The traditional one-way teaching method is ineffective; innovative approaches are necessary to stimulate their curiosity and comprehension.  

I. Background and Issues: Originating from daily life, focusing on character development  

During the implementation of the campus flower box planting program, the school has found that young children, especially those in the first grade and kindergarten stage, generally have a vague understanding of common vegetables, and they show little interest or even resistance to unfamiliar vegetables. This not only affects the effectiveness of food education, but also limits children’s ability to perceive and connect with natural life. Traditional natural education methods, such as using pictures and verbal explanations, are not effective for young children; they fail to promote deep understanding and emotional engagement.  

In response to this real issue, Shenzhen Science Research Institute of Education, Experimental Primary School (Guangming) raises a key question: How can we use human-machine collaboration strategies to organize an intelligent interactive exhibition that will encourage kindergarten children to learn more about vegetables? The goal of this project is to:  

(1) Addressing the issue of weak cognition  

By using multiple senses and engaging methods, the system helps young children better understand the characteristics and cultural significance of vegetables.  

(2) Innovative teaching methods  

Explore project-based learning (PBL) enhanced by AI technology; transform traditional learning methods into an active, creative approach rather than passive reception.  

(3) Cultivating core competencies  

In the process of solving real-world problems, students’ scientific inquiry abilities (intellectual development), artistic expression skills (aesthetic education), technical application abilities, cooperative communication skills (moral education), and practical innovation abilities (labor education) can be developed simultaneously, thereby achieving the integration of the “five aspects of education”.  

(4) Improving digital literacy  

Let lower-grade students experience firsthand the value of AI as a “creative amplifier” and a “problem-solving tool”, thereby developing their basic technical skills and digital learning abilities.  

II. Concepts and Design: Technology empowerment, full-sensing integration  

This project adheres to the concept that “life is a course, and technology is a brush”. It transforms real problems encountered in campus life into learning opportunities. By making full use of AI tools and open-source resources available within the national smart education ecosystem – such as DouBao AI and SuNo AI – a student-centered, deeply integrated smart learning model is established.  

(1) Core design concept  

1. Role transformation (Learner to Creator & Facilitator): Students shift from being mere knowledge recipients to becoming project planners, content creators, and knowledge disseminators (i.e., small mentor curators).  

2. AI-powered Multi-sensory Exploration: Utilizing AI technology to overcome traditional cognitive limitations, thereby creating immersive learning experiences that combine visual elements (AI-generated images, IPs, videos), auditory elements (AI-composed music, voiceovers), and tactile elements (interactive games, experiential devices).  

3. Human-AI Collaboration: As a powerful auxiliary tool, AI reduces the barriers to artistic creation, data processing, and solution optimization. It empowers students to express their creativity and solve problems, rather than replacing their own thinking.  

4. Cross-age Learning Community: The outcomes of this project are ultimately beneficial for the children in kindergarten. This creates a closed loop of cross-age learning: “learning in the first grade of elementary school → creation → exhibition, and experiencing it in kindergarten → feedback”, thereby promoting the social and emotional development of both groups.  

(II) Project process: Five stages of progression  

Project framework:

1. Demand research (cognitive decoding actions): Through on-site visits to kindergartens and picture questionnaires, and by using AI to analyze data, we can accurately understand the current state of vegetable knowledge and interests of target users (children in kindergartens).  

2. Curatorial design (Low-age-friendly exhibition challenge): Based on the short duration of attention of young children and their preference for tactile/physical learning (60%), students used methods such as brainstorming and group discussions. The goal was to organize an innovative exhibition about vegetables, so that the works created by the students can be displayed, thereby stimulating young children’s interest in vegetables and promoting their cognitive development.  

3. Practical validation (AI Vegetable Magic Workshop): Students use AI tools to transform ideas into practical applications:  

(1) Cognitive deconstruction: multi-sensory observation + AI interactive Q&A (Doubao AI).  

(2) Visual creativity: Doubao AI generates vegetable spirit IPs, emoji packs, and clothing design sketches; AI also converts the hand-drawn picture book “Vegetable Resource Battle” into dynamic videos (by Doubao AI).  

(3) Auditory creativity: Students created lyrics, and Su Nuo’s AI generated the theme song and MV for “Vegetable Carnival”.  

(4) Interactive experiences: DouBao AI helps to modify the code of the “Vegetable 2048” game, and creates interactive elements such as vegetable puzzles and nose-patching games.  

(5) Creative accessories: AI-assisted design of creative accessories such as vegetable clay hairpins and ID cards.  

4. Cross-age implementation (AI Curator Training Camp): Utilize DeepSeek to generate personalized presentation frameworks. Through the “Little Mentor Development Hierarchy” model (Observer – Collaborator – Leader), role-playing training is carried out (in classrooms, multimedia classrooms, and kindergarten multi-functional halls). This helps students develop their abilities to observe, express themselves, and respond effectively as “Little Mentor Curators”.  

5. Iterative improvement (The Theory of the Smart Garden’s Evolution): Conduct another survey among kindergartens, and compare the data changes before and after the exhibition; students use the “My Super Learning Skills Map” to visually reflect on their personal growth; the project is optimized based on feedback.  

(3) Technical support and resource integration  

1. Online resources: DouBao AI (key tools: analysis of vegetable characteristics, image generation, optimization of emoticons, video creation, game generation, and proposal suggestions); Sunuo AI (generation of theme songs); DeepSeek (generation of presentation scripts and optimization of proposals); open-source game code.  

Offline environments: Regular classrooms (for planning and discussion), multimedia classrooms (for digital creation and simulation exercises), kindergarten multi-functional halls (for real-life exhibitions), and campus planters (as sources of real-world scenarios).  

Learning support materials: Including series such as “AI-assisted cooperative learning sheets”, “AI-assisted project learning observation forms”, “Guidance sheets for creating songs about vegetables”, and “Guidance sheets for creating games”. These materials are created by teachers themselves, and are further improved through AI optimization. This helps to reduce the learning barriers and guide students in the direction of exploration.  

III. Implementation and Innovation: AI “paintbrush” creates amazing experiences  

The implementation process of this project fully reflects the core characteristics of educational digital transformation: technology is deeply integrated into all aspects of teaching and learning, thereby redefining the learning experience.  

(1) AI enhances deep cognition and creative expression  

1. Beyond sensory limitations: During the “Great Discovery in the Vegetable Kingdom” activity, students not only observed, touched, and smelled vegetables, but also used the DouBao AI voice interaction to ask scientific questions such as “Why do potatoes have eyes?” to deepen their understanding.  

2. Lowering the barriers to art creation: During the “Generate vegetable spirit IP” activity, students describe in natural language what they want to create (for example, “Create a spirit with a broccoli-like head and a carrot-striped shirt, who is dancing”). Doubao AI generates vivid images in real time, allowing students to instantly experience the joy of “abstracting features into tangible characters”. As a result, many funky vegetable IPs and emoji images are created.  

3. Dynamic narrative presentation: In response to the limitations of the static picture book “The Battle for Vegetable Resources” in terms of children’s understanding, students utilized the video conversion function of Doubao AI. The key innovation lies in guiding students to create a “prompt library” (actions of characters, emotional enhancements, and instructions for scene transitions) as well as optimizing parameters such as duration and keyframe duration. Additionally, AI-generated voice lines and sound effects were added, transforming the static story into an engaging animated short, thereby effectively conveying the theme of “resource sharing”.  

4. Popularization of music creation: Students create lyrics based on their understanding of the characteristics of vegetables. Using Suino AI, they select the appropriate musical style to generate melodies, and thus create a unique theme song for “Vegetable Carnival”. They also choreograph dances, transforming their knowledge into artistic expression.  

(II) AI enhances interactive experiences and problem-solving capabilities  

Gamified learning design: Students use Doubao AI to search for and modify the code of the open-source 2048 game. They transform the concept of “number merging” into a progressive logic chain: “vegetable merging = new varieties of vegetables + a fun combination of vegetables.” They replace the original materials with PNG images of vegetables, thereby integrating knowledge transfer into an interesting game format (Vegetable 2048, puzzles, and nose-patching games). In these games, the cute images of vegetables are repeatedly used, and in a way that young children can easily accept, this helps to stimulate their interest in exploring vegetables.  

2. Enlightenment on the industrialization of creative products in the surrounding area: During the “Creative Vegetable Products” activity, students analyze the characteristics of vegetables (color, shape). They use DouBao AI to gain design inspiration, then create clay hairpins and draw emoji images, and finally have those designs refined by AI. AI helps create items such as “family registers” and “identification cards” for vegetables, effectively integrating this knowledge into everyday life.  

(3) Enhancement of AI-supported roles and capability certification  

1. Personalized explanation support: In the “Little Guide” training program, DeepSeek is used to analyze the characteristics of different students, thereby generating personalized explanation frameworks, which helps reduce anxiety in public speaking.  

2. Multimodal practical exercises: Establish a hierarchical training environment: free writing exercises in classrooms (basic level) → role-playing in multimedia classrooms (advanced level) → real-scenario exhibitions in kindergartens (practical application + observer evaluation). This approach allows students to test their ability to respond to real-world challenges, such as dealing with exhibits that are not well-received or guiding children’s interactions.  

3. Data-driven reflection: At the end of the project, by comparing the survey data of the kindergarten before and after the exhibition (for example, the number of votes for “pumpkin” increased from 0 to 4, and for “coriander root” increased from 0 to 2), students can directly see the effectiveness of the project. Using the “My Super Learning Skills Map”, students mark their own progress in various aspects such as “teamwork”, “creative expression”, and “problem-solving” through coloring, thereby achieving a higher level of metacognition and visualizing their skills.  

(4) Establish a new model for intelligent teaching and research, as well as resource development  

1. The intellectual development of teachers: In these projects, teachers are not only guides, but also learners themselves. By designing and using AI to improve various learning materials and observation sheets (such as “AI Assistance – Observation Sheets for Project-Based Learning” and “AI Assistance – Guidelines for Creating Songs About Vegetables”), teachers’ abilities in digital teaching design and technical application skills are significantly improved. Students, using these tools, achieve rich results through collaboration.  

2. Generative resource library: The project has produced a wealth of reusable digital teaching resources. These include AI-generated IP images of various vegetables, dynamic picture book videos, music videos related to vegetables, interactive game programs/ schemes, and AI-optimized learning templates. These resources provide a foundation for the development of school-based courses and the sharing of resources across regions.  

IV. Effectiveness and Impact: Literacy is visible, and growth is evident  

The “Caibao Adventures” project has achieved significant educational and social benefits, effectively demonstrating the value of innovative intelligent teaching methods.  

(1) At the student level: Coordinated development of core competencies  

1. Deepened cognition and interdisciplinary understanding: Students have acquired a comprehensive understanding of the physical properties, growth characteristics, and cultural significance of target vegetables. They have also developed a comprehensive knowledge network that integrates “nature, technology, and art”.  

2. Technical literacy and creativity improved. According to the case report, 100% of students learned to use several AI tools for inquiry and creative work and understood how people and AI can work together. They designed original characters, wrote song lyrics, adapted games and designed clothing and merchandise, producing visual, audio and mixed-media work.  

3. Improvement in practical skills and problem-solving abilities: From planning exhibitions, designing exhibits, adjusting AI functions, to handling unexpected situations on the exhibition site (such as minor equipment failures or lack of interest from children), students’ problem-solving and practical skills are honed through real-world challenges.  

4. Social and emotional skills are significantly enhanced: By serving as “young mentors and curators,” students’ empathy (designing exhibitions that suit the cognitive abilities of young children), communication skills, sense of responsibility, and resilience in dealing with feedback are greatly improved. Interaction with students of different ages also helps to develop students’ leadership abilities.  

5. Learning motivation and high self-confidence: The experience of success brought about by role transformation (from learning to teaching) and technological empowerment greatly stimulates students’ interest in learning and their self-confidence. 85% of students are able to analyze data independently and draw conclusions; 100% of students can clearly understand their own progress through the “My Super Learning Skills Map”.  

(II) At the level of target users (kindergartens): Effective stimulation of cognitive interest  

The interaction at the exhibition site was very lively; on average, each exhibit received about 5 interactions. After the exhibition, the data showed that the children in the kindergarten had a significantly increased level of interest and willingness to explore vegetables that they were not previously interested in or unfamiliar with (for specific figures, please refer to the charts mentioned earlier). The holistic sensory experience provided by visual, auditory, and tactile elements helped to overcome the limitations in attention of young children, allowing them to learn while playing.  

(3) At the teacher and school levels: Improvement of the ability in smart education  

The teaching team has enhanced its understanding of PBL and STEAM educational concepts during the design and implementation of projects. This has significantly improved their ability to integrate technology (especially AI), develop interdisciplinary courses, create realistic learning environments, and implement diverse forms of evaluation. A scalable project-based teaching model based on “technology empowerment, realistic contexts, and role transformation” has been developed, along with supporting resources such as learning scaffolds and evaluation tools.  

2. The school has developed a practical approach to leveraging easily accessible AI tools and open-source technologies to promote the digital transformation of education and innovative smart teaching at low cost and high efficiency, thereby enhancing its unique characteristics and influence.  

3. Rich and diverse outcomes:

(1) Data reports: “Data Report on Children’s Understanding of Vegetables in Kindergartens” and “Comparison Chart of Children’s Understanding Before and After the Exhibition”.  

(2) Creative works: Dynamic picture book videos such as “The Battle for Vegetable Resources”; an AI-generated MV for the “Vegetable Carnival” theme; design collections for vegetable IPs; emoji packs; and designs for vegetable clothing.  

(3) Interactive experiences: Adapted version of the “Vegetable 2048” web game, vegetable puzzle games, and interactive devices for matching vegetables with noses.  

(4) Cross-age practice-related items: Real-life videos of the kindergarten exhibition, frameworks for student explanations, and lists of suggestions for improvement.  

(5) Evaluation tools: “Project Phase Evaluation Forms”, “My Super Learning Skills Map”, and a series of AI-optimized learning sheets and observation forms.  

V. Experience and Reflection: Deepening digital intelligence, and continuing to evolve  

The successful implementation of “Caibao Adventures” has provided valuable experience for innovative teaching methods in early elementary school levels. It also points out the directions for future improvement:  

(1) Core experience  

1. The real issues are the starting point: The real problems identified from campus life and student development are the fundamental basis for the viability of the project and the level of student engagement.  

2. Technology as an enabler: Clearly define the role of digital technologies such as AI as “creative amplifiers” and “efficiency tools”. These technologies should be used to support students’ deep learning and intellectual development, rather than being used merely for technical displays or to replace critical thinking.  

3. Child-friendly design is crucial: Whether it’s the choice of technical tools (simplified instructions, user-friendly interfaces), the design of learning tasks (short duration, concrete, multi-sensory), or the methods of evaluation (visualization, gamification), all these elements must align closely with the cognitive development characteristics of young children.  

4. Role transformation boosts motivation: Assigning students real roles such as “curators”, “mentors”, and “designers” can greatly enhance their sense of responsibility and internal motivation.  

5. Closed-loop iteration promotes growth: The closed-loop design of “requirement – design – implementation – feedback – iteration” ensures the effectiveness of the project and the development of students’ reflective abilities.  

6. The integration of the five aspects of education is the ultimate goal: the program naturally combines moral qualities (sense of responsibility, cooperation), intelligence (exploration, cognition), physical ability (practical application), aesthetics (artistic creation), and creativity (transforming ideas into tangible products), thereby promoting the comprehensive development of students.  

(2) Reflection and optimization directions  

1. Improve technical compatibility and stability:

(1) Develop/select a more simplified, child-friendly AI interaction interface (such as enhanced voice commands, icon-based operations).  

(2) Strengthen the compatibility testing for previous technologies (e.g., to resolve occasional lag issues during the generation of dynamic picture books).  

(3) Explore the use of AI emotion recognition technology to analyze in real time the children’s participation and emotional responses during the exhibition, thereby providing immediate feedback and optimization suggestions for the “little mentors”.  

2. Deepen understanding and expand the curriculum:

(1) Longitudinal extension: Incorporate more in-depth topics such as the observation of the vegetable life cycle (e.g., in conjunction with campus farming practices) and research in nutrition science.  

(2) Strengthen collaboration between schools and families: Introduce “Family Vegetable Innovation Competitions” and “Parent-Child AI Workshops” to expand the impact of these programs to families, thereby promoting joint education between schools and families.  

3. Improve process evaluation and data collection:

(1) Use AI tools to automatically record the creative processes of students (such as operation screenshots and voice logs), thereby capturing their development trajectory more comprehensively.  

(2) Explore the use of AI image analysis techniques to quantitatively analyze the attention distribution and interaction frequency of children during exhibitions. This will provide more objective feedback data, which can help optimize the design of exhibits.  

(3) Further refine the evaluation dimensions and grade descriptions of the “Super Science and Technology Skills Map”.  

4. Promote resource sharing and regional influence: Systematically organize the high-quality resources generated by the projects (learning materials, AI prompt libraries, examples of works, teaching models). These resources can be shared through regional education platforms or the national smart education platform, thereby benefiting more schools and teachers.  

6. Conclusion  

“Caibao Adventures” is a remarkable achievement made by Shenzhen Science Research Institute Experimental Primary School (Guangming) in the context of the digital transformation of education. Based on real problems encountered in school life, and using AI tools from the national smart education ecosystem as tools, along with project-based learning methods, this initiative successfully helped first-grade students embark on an intelligent and holistic exploration journey, from “cognitive decoding” to “creative exhibition”. It demonstrates how digital technology can greatly enhance the integration of the five aspects of education, how it can effectively address the challenges related to abstract thinking and deep learning in early childhood, and how it can stimulate students’ unlimited creativity and practical wisdom. The concepts and practices outlined in this project—such as “life is the curriculum, technology is the tool”, “learners are creators and disseminators”, and “AI as a tool for creativity”—provide a valuable “Shenzhen example” for innovative teaching methods and digital transformation in primary schools. We will continue to focus on refining these approaches in terms of technical adaptability, course depth, evaluation accuracy, and resource sharing, so that every child can explore real-world problems in this digital age and develop healthily and happily.