Overview and Inspirations
Inspired by the belief that knowledge is the catalyst for change, our education outreach initiative is designed to bridge the scientific community and the public for a sustainable future. In an era with complex global challenges like plastic pollution, we cannot allow innovative solutions to remain confined to laboratories and academic papers. Through a variety of approaches including interactive sharing sessions, hands-on project exhibitions, data science experience courses, clean recycling events, tailored educational booklets, and surveys to collect feedback and suggestions about the educational booklet and project exhibition, as well as board game design, we not only disseminate knowledge but also listen to public concerns and values through two-way dialogue.
Our approaches are tailored to resonate with a wide spectrum of learners, including primary school students, secondary school students, parents, and the general public. Our mission is to translate cutting-edge science and technology into accessible, engaging, and actionable knowledge, empowering everyone to be part of the solution. The greatest challenge and most important goal is to make complex concepts simple without being simplistic. We break down the key messages as follows:
- How does our engineered biosensor work?
- How to raise public awareness about environmental justice, a concept that is not yet widely known?
- How do the AI models Mask-RCNN and YOLOv8 work? How do these AI models detect and quantify microplastics?
- How to achieve clean recycling?
Sharing Sessions in Primary and Secondary Schools
Our team conducted various sharing sessions at primary and secondary schools, reaching 600 F.1 to F.5 students in G.T. (Ellen Yeung) College Secondary Section on 7th July; 300 P.4 to P.6 students in G.T. (Ellen Yeung) College Primary Section on 8th July; 600 F.1 to F.5 students from Queen Elizabeth School on 9th July; and 287 P.3, P.4, and P.6 students from Salesian Yip Hon Primary School.
Inspired by suggestions from Dr. David Jones, we adopted flexibility in our presentation approach. We adjusted our presentations to different age groups, ensuring a diverse range of audiences could engage. For primary school students, we made the sharing sessions more fun, interactive, and easy to understand.
To simplify the concept of biosensors, we used an analogy to explain what a biosensor is and how it can be used to detect microplastics. The analogy was as follows: We compared a biosensor to a typical Lego Mindstorm piece. On the screen, there were two Lego sensor pieces and a light bulb. The Lego sensor on the left side represented LasR in the sensory module, while the Lego sensor on the right side represented pLasRL in the reporting module. They were connected to a light bulb (hydrogel of X-gal assay). When a beam of light (AHL molecule secreted from bacteria adhering to microplastics) entered the Lego sensor on the left side, it (the LasR-AHL complex) transmitted a signal to the Lego sensor on the right side, which converted the signal into a current, causing the light bulb to light up (resulting in blue coloration in the hydrogel).
Secondly, we modified our presentation by using larger, colorful pictures and breaking down the concepts into separate slides. For example, we used colorful shapes to represent the basic components of the BioBrick circuit and taught students the sequence of the basic BioBrick circuit: Promoter → Ribosome Binding Site → Coding Sequence → Terminator. We then invited two groups of four students each to come on stage. Each student was given a cardboard with one of the four components of the BioBrick circuit. They had to arrange the components in the correct sequence of the circuit. The students actively participated in the game and were able to arrange the correct sequence successfully.
We brought up the idea of environmental justice with some examples, such as the illegal plastic recycling factories in Jenjarom and the Flint water crisis. Then we used an innovative way to encourage active participation of students, which is an interactive video about environmental justice. Students were fully engaged in a choice-driven video in which they were told to be a boss of a restaurant and have to make choices on several decision such as what packaging of straw, plastic or paper to choose and how to dispose of plastic waste, dump into Less Developed Country which is cheaper or go recycle which is more expensive. They create a unique story and ending to the video. Students can raise up their hands to decide how the story goes on, and different choices lead to different endings. We hope to make new and difficult topics much easier for the students to understand by letting them really participate as a character in the story.
To make students understand the working principles of AI models, we used “Where is Wally”, a series of children's books where the goal is to find a distinctively dressed character, Wally hidden within highly detailed illustrations to explain the one step and two steps process of AI models, Mask-RCNN and YOLOv8. We also used robot performance to show the comparison results of three AI model that our team used in building the suitable model installed in the robotic car. Robot with the best score represent the best AI model performance which is YOLOv8 in our project.
We also educated students how to clean recycle beverage carton by live demonstration during sharing session. In all sharing sessions, students were fully engaged and gave very positive responses.
Data Science Experience Course for S1 to S5 students
We organized Data Science Experience Course on 23rd April, a course designed by our Technology Team members. The course aimed to allow students to learn more about the AI models, Mask-RCNN, RT-DETR and YOLO v8, and have a hands-on experience about plastic tagging. We hope the activity will allow students to acquire practical skills on training an AI Model and how to make good use of large datasets.
The course began by contrasting the traditional, manual method of microplastic collection with the potential of AI-driven automation, followed by an introduction to the viable AI models (Mask RCNN, YOLOv8, and RT-DETR) for detecting microplastics. We also displayed some work from previous research and what is being done by the Technology Team and showed the results for the images of the microplastics after they were processed by the AI models.
The course was a great testing ground for us; we identified key areas for improvement. Our content was too technical, with terms such as 'RolAlign' and 'FCN', which were proven to be too advanced for participants. Thus, we made our content less in-depth and using more analogies in subsequent courses in September, which were less demanding without removing the core contents and key points we wanted to deliver. It was proved successful, as the participants gained a good understanding of the use of AI in microplastic detection and the integration of synthetic biology with AI.
Benefiting from the experience we gained and our overall growth in knowledge during the summer, we revamped our presentation by delivering complex ideas through analogies, such as using the popular "Where's Wally?" game to demystify how AI models like Mask-RCNN and YOLOv8 perform object detection. We also explained the processes of developing an effective AI model (train, verify, test) by using the way students prepare for exams. Besides, there was a plastic tagging activity to teach students how to use Roboflow annotation tool to tag the microplastics taken by our team.
Furthermore, we split the event into 3 days, one for Grade 7 students, one for grade 8 students, and the final day for grade 9 to 11 students. This arrangement allowed us to adjust our difficulties based on students’ academic level and maturity. This was especially beneficial for explaining machine learning where mathematics abilities are involved. For example, the equation y=mx+c may be relatable to gradient descent for G9 to 11 students.
We conducted pre-surveys and post-surveys to study the responses of participants. The surveys showed strong evidence that participants had a substantial increase in their understanding of our contents. From the first question, it can also be concluded that participants, after the course, have a strong understanding of the fundamental regarding the AI models as the number of students understand the working principle of the two models increased from 55.6% to 88.6%.
Pre-survey
Post-survey
At the end of the course, we collected valuable feedback together with the post-survey. Nearly all participants strongly agree or agree that they learned more about different AI models’ algorithms. Also more than 95% participants strongly agree and agree that they understand how plastic tagging could be done. Together with the performance on the post-survey, indicates the event was a success.