CS education for all has been established as an important area of focus for both researchers and practitioners in recent years. At the same time, due to the increasing prevalence of AI technology in humans' everyday lives, arti cial intelligence (AI) education for K-12 is gaining special attention among CS educators. AI literacy, even more than general CS competencies, requires evidence-based research to be e ectively integrated in our schools. The common learning environments utilized for CS education enable us to go beyond conventional educational research approaches by providing a platform where detailed data can be collected from students' interaction with CS-education-related activities. Thus, conventional educational research approaches coupled with insights gained from pattern recognition and student modeling approaches enable us to e ectively improve our instruction and to provide students with adaptive sca olding. In this work, we present our rst AI curriculum module that is designed to teach a fundamental AI search algorithm, Breadth-First Search (BFS), through a series of progressively sca olded activities. Data is collected from a preliminary pilot of this activity with a high-school student in the form of a think-aloud protocol, screen capture, submitted block-based programming artifacts, and interview questions. Our results demonstrate that our activities have been successful in increasing the student 's knowledge about the BFS algorithm and more importantly, how this particular AI algorithm can be utilized to solve real world problems. Based on the results of this pilot study, we propose designing a comprehensive AI curriculum contextualized within a learning environment that collects detailed data from students' progress to inform instructional design and facilitate adaptive sca olding for students.
Copyright ©2021 for this paper by its authors. Use permitted under
Creative Commons License Attribution 4.0 International (CC BY 4.0)
Integrating AI education into the K-12 curriculum poses
signi cant challenges for educators since teachers often lack
prior education in CS related elds [
Section 2 discusses related work. The details of the devised curriculum are presented in section 3. The case study is presented in section 4, and the results of the pilot study are discussed in section 5. Section 6 concludes the work, and section 7 discusses implications for design and future work.
Emerging research is exploring the design of learning
experiences to foster youths' Arti cial Intelligence (AI)
literacy so that they are prepared to enter and engage with an
AI- lled future [
We developed 5 exercises in UC Berkeley's Snap block-based programming language that introduce and use a simple graphical interface for teaching the breadth- rst search algorithm (BFS). The rst four exercises serve as an introduction to BFS, and feature a character, West, pictured in Figure 1, who guides student through the steps of the algorithm by telling them how to act as a GPS. The last activity is formatted as a programming exercise, where students are given an incomplete implementation of BFS, and are required to ll in the gaps. The algorithm is used for contact tracing, similar to the COVID contact tracing apps, to showcase a real-world application of BFS.
Exercise 1 is designed to introduce graph search as a problem in the context of a GPS navigating between cities, as well as to show how the methodologies for performing this search di er between computers and humans. West shows the student a city map as a graph, with nodes representing cities and transitions representing roads between them (Figure 2). The student is then instructed to click through the cities to nd the shortest path from a starting city to a goal. The student does this rst with the simple map shown in gure 2, then with a more complicated map, and then nally with a map which has cities hidden from the student, which can only be shown once they click an adjacent city. This last example is meant to describe the perspective of the algorithm to the students, to contrast it with the human perspective of seeing the whole state space at once.
Exercise 2 is designed to introduce BFS as an algorithm for performing graph search. BFS is introduced to the students via a series of animations and they are instructed to perform each of the steps of BFS by interacting with our graphical interface, shown in Figure 3. This interface provides two ways to perform the actions of the algorithm. The rst is to add nodes to the fringe by clicking them on the displayed graph. The second is to remove nodes from the fringe. The legend in Figure 3, which shows the graphical representations of each state for the nodes and transitions, is displayed and the interface updates accordingly as students interact with it.
Exercise 3 consists of a completely guided run-through of BFS for a medium-sized graph with eight nodes and ten transitions. At every step of the algorithm, students are told what to do next and upon making a mistake, they are informed about what the mistake is and re-prompted until choosing the correct option.
Exercise 4 is composed of a partially guided run-through of BFS. To connect this exercise with a real-world application, we show the graph superimposed over a map of the east coast of the U.S., with the nodes named for their corresponding cities. For the duration of the exercise, students complete the steps of the algorithm without feedback unless they make a mistake. Upon making a mistake, they are informed about the nature of their mistake and re-prompted until they perform the correct next step.
Exercise 5 challenges students to implement their understanding of BFS as Snap code. For this exercise, we recontextualized the state space for BFS from a map of cities to a social network of people. When the exercise is initialized, one to three random people are marked as infected with COVID-19. The objective for BFS was changed from nding the shortest path from a starting city to a goal city to nding the shortest number of social connections from a user-selected person to an infected person. This scenario was inspired by the COVID contact tracing apps that have recently been developed and implemented in order to show students a real-world application of BFS. Figure 4, right shows the graph of people with the infected people in red. Students were given a partial implementation of the algorithm along with missing blocks, with some instructions on how to snap the blocks into their proper place (Figure 4, left). Upon running the algorithm, if it is correctly implemented, students see the path from the user-selected person to the nearest infected person.
This study employed the case study method [
The case study participant was Selim (pseudonym). He was a student in the 11th grade at a public high school in the United States. He had an interest in Arti cial Intelligence and was actively looking for opportunities to learn about how AI worked and how AI could be used in the eld of his interest, Biomedical Engineering. He had some general knowledge about AI and Python programming. The study took place in the context of two informal virtual sessions that were designed to introduce Arti cial Intelligence with the BFS search algorithm. The sessions were conducted remotely, a week apart, via Zoom, a video conferencing tool. The rst session took approximately one hour. In this session, we rst described the goal of the session and emphasized that the student was expected to experience the rst four learning activities as a user, complete a post-activity questionnaire, and participate in an interview. The second session was mainly focused on the fth activity and took 1.5 hours. In this session, Selim answered questions related to the previous session, followed a written Snap tutorial, completed the activity, and participated in the interview.
In this study, we collected data in the form of a thinkaloud protocol with approximately thirty minutes of a semistructured interview, post-questionnaire, screen capture, and submitted block-based programming artifacts. In the rst session, we asked the participant to answer warm-up questions (e.g., Have you ever taken any class on AI?) and complete an activity that aimed to extract his interests and general knowledge about AI. In addition to the warm-up activity, at the end of each learning activity, we also asked \What do you think you have learned? What else would you like to explore?". With these questions, the participant reected upon his own learning. Also, after he completed the activities, we asked him to complete the post-questionnaire that included seven multiple-choice questions related to the BFS learning activities. At the end of the rst session, we conducted semi-structured interviews for approximately 10 minutes to learn more about his experience. In the second session, we followed a similar structure. We started the session with the warm-up questions to recall the student's understanding from the previous session and to correct misconceptions. Before the activity, we guided the student through the creation of pseudocode which re ected his understanding of the algorithm. We then provided a written Snap tutorial to allow the student to get familiar with the learning environment. After he skimmed through the tutorial document, he moved on to the fth activity. At the end of the activity, we conducted an interview to learn more about his experiences.
To analyze the transcripts of interviews, we followed the thematic coding method [2], which is a widely used method to analyze qualitative data for identifying, organizing, and describing themes. In the rst round of the analysis, one researcher analyzed the data to explore learning challenges and opportunities. After the rst-round coding, a peerdebrie ng meeting was held with three other researchers to ensure the trustworthiness of the ndings. Additionally, we analyzed the student's process of building the blockbased programming artifacts by following the thematic coding strategies. We open-coded the students' actions in the activities to identify learning challenges and opportunities. After that, we discussed these codes and came up with the general themes discussed in section 5.3.
Results are organized around three overarching themes: teaching AI as an interdisciplinary eld, creating opportunities for students to discuss who bene ts from AI and is left out, and promoting e ective AI learning through developing a humanized curriculum.
We should provide opportunities for students to understand the integration of AI in application areas and teach AI as an interdisciplinary eld. Selim was interested in the eld of biomedical engineering and believed that AI would contribute to innovations in every eld, including biomedical engineering. Thus, he was motivated to learn more about the application of AI in the eld of his interest. During the rst session, he shared, \If you are trying to determine, for tissue engineering, maybe you could use a model to determine what's the best material to use based on di erent situations." At the end of the rst session, Selim asked the research team about learning experiences in biomedical engineering and computer science in higher education since he was at the critical stage of choosing future careers. His question highlighted the need of supporting students to develop interdisciplinary learning and collaboration skills and understand the integration of AI in elds of their interest.
Our analysis indicates that topics related to AI ethics might serve as a catalyst for students to have in-depth discussions about who bene ts from AI and who is left out. For instance, in the excerpt below (Excerpt 1), guided by the research team, Selim emphasized that AI ethics could be an engaging topic of discussion. Excerpt 1.
1. Research team: Have you talked about AI with your family and friends? 2. Selim: Not really my friends. With family, yeah. We've had a few discussions. Um, it's probably a really common discussion. I think one of my parents heard it on a podcast or something like the thing about a selfdriving car having an unavoidable accident. 3. Research team: Yeah, I remember this one. 4. Selim: Like, who it (referring to the self-driving car) picked to crash into or something. Just like ethics of AI is probably the most interesting conversation to have with other people.
In Turn 4, he highlighted that a model would fall short in addressing the moral question of who a self-driving car should kill in an unavoidable accident. However, he did not recognize that essentially, it's the car maker or designer who would make the decision.
1. Research team:Do you nd AI ethics interesting? 2. Selim: Oh, yeah. 3. Research team: In what ways? Can you give an example? 4. Selim: So another thing that I found interesting is the hiring process. I think it's Amazon. It has implicit bias against hiring applicants that had women in their application. That's obviously an important issue that needs to be xed. So there are really complex issues when you talk about AI ethics.
In turn 8, he raised the concern about gender bias using AI recruiting or resume screening tools. Overall, we can see that AI ethics and AI decision making that would have an impact on who bene ts and is left is a promising topic for students to think critically about the impacts of AI technologies.
As described in section 3, the student is presented with a series of ve activities. The rst four activities are designed to introduce the BFS algorithm to the student while the fth requires him to implement the BFS algorithm in a blockbased programming environment in the context of a contact tracing app. Selim progressed though the rst four activities rather smoothly. After each activity he indicated a deeper understanding of the BFS algorithm. He then successfully answered content knowledge questions presented to him at the end of the rst session In the beginning of the second session, the student started by re ecting on what he recalled from the previous session and was then asked to devise pseudocode for the BFS algorithm. Through this process he demonstrated a good understanding of the BFS algorithm but a lack of familiarity with writing pseudocode. With the guidance of researchers he managed to turn his knowledge of the BFS algorithm into a pseudocode format. He was then presented with an incomplete contact tracing app that utilizes BFS to nd the shortest path between an arbitrary person and an infected person in his social circle.The BFS algorithm was partially implemented as a Parson's problem, and the student snapped the readily available blocks on the screen to complete the algorithm. Here again, the student showed an adequate understanding of the BFS algorithm. However, he faced some challenges when trying to implement it within the blockbased programming environment. We identi ed three main themes of challenges faced by the student while conducting the programming. First, he was not familiar with some of the blocks, list-related blocks in particular, and was confused about what they represent in the incomplete program. Secondly, he had trouble understanding the functionality of some of the custom blocks provided for him. Finally, the student was unclear about how to map from steps of his pseudocode onto the di erent sections of the partial code. While the rst challenge can be addressed by providing students with a more extensive block-based programming tutorial beforehand, the second two challenges require improvements in the design of the activity. For example, a better description of the functionality of custom blocks and a clearer distinction between custom blocks and contextually named variables can alleviate some of the confusions that students might encounter when analyzing the partial code presented to them. Additionally, having students implement a simple case of BFS following the same context as the previous activities might facilitate the transition to implementing the BFS algorithm in an arbitrary real-world context.
With the advances of AI technology and its rapid integration within our society, it is imperative that AI education is included as part of the K-12 standard curriculum. In this study, we presented the preliminary design of a curricular module that is designed to teach a foundational AI algorithm, breadth- rst search (BFS), to high school students. We presented the results of a pilot implementation of this curriculum with a high-school student as a case study. Our results demonstrated that this intervention helped the student gain a better perception about the interdisciplinary nature of AI and how it can be incorporated to enrich other elds. Furthermore, the discussions fostered curiosity around AI ethics which is an important aspect of AI education. Finally, our analysis of the student's interactions with the designed activities and his re ections showed that the activities were successful in improving the student's understanding of the BFS algorithm while posing some challenges when the student tried to map from the algorithm to a block-based programming implementation.
In the future, we aim to utilize insights obtained from this case study to not only improve the design of the current module but also to inform design of the future AI-related curricular modules. While this study provided a qualitative analysis of observed interactions between the student and activities, we plan to automate this process in the future through data-driven approaches to obtain more insights into the e ectiveness of our curriculum and students' learning. We propose creating a series of curricular modules that teach students a holistic view of AI and equip them with knowledge, skills, and abilities that prepare them for conducting problem-solving with AI in humanized real-world scenarios. We further propose situating our curriculum within a learning environment where data can be collected and analyzed for informing instruction and providing students with adaptive sca olding. 8. REFERENCES [1] M. Bassey. Case study research. Research methods in educational leadership and management, pages 108{121, 2002. [2] V. Braun and V. Clarke. Using thematic analysis in psychology. Qualitative research in psychology, 3(2):77{101, 2006.