: AI in CS1: Supporting Learning Without Replacing Thinking

Speaker: Michal Shemesh (Ben-Gurion University of the Negev) Title: AI in CS1: Supporting Learning Without Replacing Thinking Abstract: How can we harness the power of AI to support students' learning without compromising their ability to think and solve problems independently? In this session, I will present three projects being implemented this semester in the Introduction to Computer Science course for Computer Science and Software Engineering students. Together, these projects address a central educational challenge: how to take advantage of AI tools while cultivating the independent thinking and problem-solving skills that are fundamental to computer science education. The first project - AI-Assisted Learning The first project is a pilot of a system developed by the Faculty's IT team, led by Vitaly Shapira and Mark Kappel. Students work in a closed VS Code environment alongside Maia, a friendly AI agent that accompanies them as they complete their homework assignments. A key feature of the system is the lecturer's ability to define and control the agent's behavior, determining how it guides students, responds to their questions, and supports their learning without taking over the problem-solving process. I will present the principles and specifications underlying Maia's design, the process that led to its development, and a live demonstration from both the student's and the instructor's perspectives. The second project - Independent Problem Solving While AI assistance can enhance learning, the ability to confront unfamiliar problems, struggle with them, and develop solutions independently remains essential. This is precisely the ability we strive to cultivate in our students. The second project provides students with opportunities to practice solving problems as they would on a whiteboard, without an IDE, an AI agent, or any other external assistance. This training is designed to complement their AI-assisted programming experience, strengthening their confidence and ability to think independently. I will share initial results and student feedback. The third project - AI-Based Assessment The third project builds on the AI-based homework grading system developed last year and presented at this seminar. We have extended this work to support the assessment of students' independent, handwritten problem-solving assignments. Beyond its application in our course, the system opens possibilities for automatically grading handwritten, non-programming assignments in a wide range of academic disciplines. I will discuss the system's capabilities and its potential applications. Together, these three projects represent a complementary approach to AI integration in education: using AI to guide students when assistance is valuable, encouraging them to think independently when assistance is deliberately removed, and using AI to assess their work and support the learning process. The goal is not simply to teach students to work with AI, but to ensure that they also learn to think without it. Time and interest permitting, I will briefly present how we chose to channel the human resource, the most valuable resource in the course, in a way that has brought, and continues to bring, students into the classroom, even though all class sessions are recorded and published, and all course materials are available in advance on the course website with unrestricted access.