In this post, we feature Mr Michael Tan Yong Heng, a Senior Lecturer from the Division of Information Technology & Operations Management at the Nanyang Business School. Mr Tan has been teaching Artificial Intelligence (AI) to students without a computing background, and he shares with us his innovative approach to make this complex subject accessible and engaging for his learners. In his lesson vignette, Mr Tan reveals how he strategically employs technology and peer learning to optimise class time and prioritise the development of essential thinking skills and learning dispositions. His ultimate goal is to equip his students with the necessary tools to navigate and thrive in a world increasingly shaped by AI and technology.
Teaching Approach
Learning anything new is never easy. I believe that for students to achieve deep understanding takes time, especially to appreciate abstract concepts like the workings of Artificial Intelligence (AI). Unlike AI, the human brain can easily be overwhelmed when presented with complex concepts. Thus, as an instructor, I am cognizant of the need to design learning experiences that help my students better manage their cognitive load so that they can remember and assimilate new information with what they already know.
Scaffolding students’ knowledge building process by giving sufficient focus on the basics at the start to lay down a strong foundation before moving on to more complex subjects is critical. This way, students don’t feel swamped at the onset, and gain confidence to get ready to tackle tougher topics. It’s like doing a puzzle: you start with the edge pieces to build the frame, and then you fill in the middle bits to see the whole picture.
In my classes, the pace at which we move depends on how well the students are picking up the material. Their understanding is more important to me than just sticking to the lesson plan. This flexibility means I can tailor my teaching to fit their needs, ensuring nobody gets left behind. Teaching here is very much a team effort, with plenty of chances for feedback and discussion through informal formative assessment, helping me fine-tune how fast or slow we go based on how well the students are doing.
My teaching is crucial to keeping students engaged, making the material relevant to their lives, and providing hands-on learning opportunities, which aligns with the principles of active and authentic learning. I work to make lessons interesting with lively discussions, examples from the real world, and group projects. Linking what we learn to real-life situations makes the subjects more appealing and relatable, encouraging learning and sparking curiosity.
Also, I favour doing things over just talking about them. Students can put what they’ve learned into action with practical exercises, simulations, and teamwork. This deepens their understanding and boosts their ability to think critically and solve problems. I applied these principles in one of my classes on Artificial Intelligence.
Making Artificial Intelligence Understandable
Teaching the course CC0002 Navigating the Digital World means I often work with students unfamiliar with computing. To explain Artificial Intelligence, I avoid technical jargon and use simple language and analogies familiar to students. Also, I break down complex ideas into smaller parts and use visuals to make abstract concepts more straightforward, applying the concept of dual coding, whereby words and visuals are combined to enhance learner understanding.
From Traditional Programming to Machine Learning
I start with traditional programming, showing how it differs from machine learning through basic examples. For instance, traditional if-then-else statements can’t always accurately identify activities like playing golf.
This leads to a discussion on Machine Learning, where I explain how computers can recognise activities by learning from data labelled as “walking,” “running,” “biking,” or “golfing.”
Understanding Supervised Learning
After covering supervised learning, I use the example of predicting temperatures to introduce students to selecting the best model through trial and error. This illustrates how computers can learn from data to make predictions.
Hands-On Learning with Legos
One powerful instructional approach I employ is concrete representation, which ties into the principle of embodied cognition. Concrete representation involves using physical objects or visual aids to represent abstract concepts, making them more tangible and easier to grasp. In one lesson, I used Legos to teach the concept of k-means clustering, a machine learning algorithm that groups similar data points together.
I started by arranging 10 light blue Lego bricks in a line to represent data points, telling students they could imagine these represented images of cats and dogs. I then placed a yellow Lego on one light blue brick and a dark blue Lego on another, explaining these represented the initial centroids (i.e., the center of each cluster).
Students took turns sorting the remaining bricks into the yellow or dark blue cluster based on which centroid they were closer to. We measured distance simply by counting the bricks between each point and the centroids. After sorting all the bricks, we evaluated our clusters by seeing how spread out the bricks in each one were, again just by counting. This roughly represented the mathematical concept of variance used to evaluate clustering accuracy.
Our first attempt did not cluster the data points correctly, so we repeated the process, moving the centroids and re-sorting. This illustrated the iterative refinement process of machine learning. Students experimented with different centroid positions to try to create the tightest groupings, engaging in trial-and-error learning. We discussed the results of each attempt and why certain arrangements worked better.
Throughout this activity, students were actively involved, both physically manipulating the Legos and engaging in discussion and problem-solving. The familiar objects and lack of complicated math kept them engaged. Afterwards, students expressed a much clearer understanding of what k-means clustering actually does and how it works through repeatedly refining its clusters.
Concrete representation is a powerful instructional approach for teaching complex topics like machine learning to students without a technical background. By using simple, relatable objects like Legos to physically demonstrate abstract algorithms and processes, we can foster hands-on learning, engagement, and conceptual understanding.
This approach, embedded within the broader context of a carefully paced and conceptually grounded curriculum, cultivates a learning environment where students can connect with the material in a meaningful way.
Conclusion
In reflecting on the teaching journey and the methodologies employed in demystifying Artificial Intelligence for students without a computing background, the core principles of machine learning—trial and error and the significance of data for predictions—stand out as pivotal learning moments. Through the practical exercises of predicting temperatures and clustering LEGO bricks, students not only engage with the material in a hands-on manner but also internalise two fundamental truths about machine learning.
Firstly, the temperature prediction activity underscores the iterative nature of machine learning algorithms. It exemplifies how, in the realm of artificial intelligence, precision is the product of persistence. The concept of using trial and error to refine predictions enables students to appreciate the algorithmic process of learning from mistakes, mirroring the very human process of learning. This parallel between machine learning and human learning processes reinforces the idea that at the heart of AI’s complexity lies a very intuitive principle: learning from experience.
Secondly, the LEGO clustering activity brings to light the crucial role of data in shaping AI’s capabilities. By categorising LEGO bricks without predefined labels, students grasp the essence of unsupervised learning and understand that the quality and volume of data directly influence the algorithm’s ability to discern patterns and make informed predictions.
Together, these activities illustrate the practical applications of machine learning and embody AI’s iterative, data-driven nature. This hands-on approach, employing relevant elements of the science of learning, embedded within the broader context of a carefully paced and conceptually grounded curriculum, cultivates a learning environment where students can see beyond the abstract and technical, connecting with the material in a meaningful way.
