Ben-Gurion University Pilots AI Code Assist to Guide Computer Science Students
Ben-Gurion University is launching a pilot for AI Code Assist, a custom system designed to guide 600 computer science students with hints instead of providing ready-made code.
Ben-Gurion University of the Negev is tackling one of the major challenges artificial intelligence poses to academia: students receiving ready-made solutions to assignments while missing out on critical hands-on experience. In the upcoming semester, approximately 600 students will participate in a pilot program for a system called AI Code Assist, designed to guide them through programming exercises using hints and leading questions rather than writing the code for them.
Developed by the Department of Computer Science, the system will be integrated into the Introduction to Computer Science course. Teaching teams note that homework assignments play a central role in acquiring core skills such as writing code, debugging, and independently solving problems. According to student data provided by the university, practical exercises previously accounted for roughly 75% of their learning time throughout the semester.
Curbing Over-Reliance on External AI
However, the widespread availability of generative AI tools has altered this dynamic, allowing students to obtain completed code before learning how to break down complex problems or identify why their solutions fail. Teaching staff reported a decline in deep comprehension over the past year, which was also reflected in final exams.
Mark Kappel, a teaching and research laboratory engineer leading the development alongside the IT division, explained:
"Our development creates a controlled and closed working environment that prevents the use of external AI tools, allowing the teaching staff to control the nature of the system's responses under the hood."...
A Digital Tutor Instead of a Shortcut
Rather than banning artificial intelligence, the course instructors aim to reshape how it is utilized. Lecturers Asra'a Nassasra, Prof. Assaf Zaritsky, and Michal Shemesh noted that the system transforms AI from a tool that generates ready-made answers into a digital tutor that enforces critical thinking.
According to Nassasra, the teaching staff will predetermine the level of assistance permitted. The system provides tiered hints, suggests breaking problems into smaller steps, and asks guiding questions to highlight logic flaws without providing the final answer.
Future Expansion and Assessment
Shemesh emphasizes that the challenge is particularly acute for first-year students who are still building their foundational knowledge base. Unrestricted AI usage bypasses trial-and-error processes vital for long-term academic success.
The university plans to evaluate the pilot across two versions of the introductory course involving Assaf Zaritsky and Michal Shemesh, with potential expansion into advanced courses and other academic departments in the future.
