. Unplugged activities can support the development of student interest and engagement, while teaching a wide range of AI concepts [4]. For example, Ma et al. [5] introduced two unplugged activities, "The Pasta Land-Decision Tree" and "The k-NN-Penguin, " which incorporate decision tree and neighbors algorithms tailored for middle school students. These activities use familiar animals and objects to help students grasp AI concepts in a relatable and age-appropriate way.
In our prior work, we designed, developed, and piloted an unplugged AI learning activity, "Guess Whose Face" [3]. We implemented the activity with 10 rural middle school students during an AI-focused summer camp. Results from exit tickets taken at the end of the activity indicated that the students scored an average of 4.625 out of 5 for enjoyment and engagement and 4.375 out of 5 for conceptual knowledge about Representation and Reasoning. Furthermore, we noted that letting students and educators participate in creating and designing part of the unplugged activity could offer promise for further promoting students' engagement. Previous work has explored a variety of toolkits that are available as a means of providing customization opportunities for students [2]. Drawing on the significant potential of unplugged activities for K-12 AI education and informed by our prior findings, we present our work on a game construction toolkit, Card Craft Studio, which enables students and educators to customize our "Guess Whose Face" unplugged AI learning activity.
Prior to designing our toolkit, we developed an unplugged activity called "Guess Whose Face" aligned with one of AI4K12's Big Ideas in AI, Representation and Reasoning. Unlike the popular board game "Guess Who?" wherein players ask questions to guess the identity of a person, our game requires students to use features represented as dots extracted from a facial image to guess the person.
The game is designed to be played as a team of two or more students to compete against other teams. There is one Drawer and one or more Guessers on each team. The role of the Drawer is extracting facial features onto tracing paper from a randomly chosen face card without revealing the target face (Figure 1). Guessers use a spinner to determine the team's next action and have to recognize the target face from the facial features drawn on the tracing paper. The face card deck serves as a dataset of target faces that the Guessers attempt to identify. For the summer camp, we created Through our pilot study with summer camp attendees, we noted diverse student behaviors and collected insightful suggestions that can aid in refining the activity. We observed that students' participation in the activity appeared more enthusiastic when they chose a card of a person or character they knew or were familiar with. Conversely, students seemed to be less enthusiastic when choosing an unknown or unfamiliar card. We collected information about celebrities and characters middle school students prefer through a post-camp survey. The survey suggested that there was a limitation in the original design of the unplugged activity that we could not meet the preferences of all students due to the fixed deck of face cards we provided. One possible solution is to provide a printable DIY (do it yourself) template. This is a well-known method among people who play "Guess Who?" with customized card decks. Using a template supports placing an image file in a fixed number of empty squares that players can print out after mapping all of their own images. Although this method is fairly straightforward, it has the disadvantage of requiring labor-intensive effort to place the images manually. To address this problem, we designed a web-based toolkit allowing students and educators to choose the desired images and automatically arrange them to print their unique face card deck.
The toolkit is designed with simplicity and intuitiveness in mind, aiming to allow first-time users to use it effortlessly (Figure 2). The design of the toolkit emphasized four core functionalities.
Card Configuration. Users can designate a color for the card based on the intended game difficulty levels (Easy, Medium, or Hard). They can also freely select any number of cards to create, ranging from 1 to 20, for customized gameplay.
Card Image Upload. This step facilitates the uploading of facial images for creating cards. Users can either 'drag and drop' an image file or choose one from an existing set of images. By default, the title of the uploaded file will be displayed as the name of the individual shown on the card; however, this can be edited.
Card Production. This feature automatically lays out the uploaded images into a card template, generating a personalized face card deck, and users have the option to print their crafted cards. Card Sharing and Editing. Users can save their facial images as a project file so that they can later reload it to modify, share with peers, and print it out for future usage.
Fostering student engagement in K-12 AI educational activities is critical. We have presented our work on a toolkit to enable students and educators to create customized versions of an AI-focused unplugged card game activity to help promote student interest in the activity. The toolkit is being developed as an open-source webbased application that will be available to educators and students interested in using our unplugged activity for K-12 AI education. In the future, it will be important to gather feedback on the toolkit usage from students and educators for future refinements. We aim to collaborate with local communities passionate about AI education, guiding them on how to use the toolkit and card game as we evaluate its effectiveness in practice.