AI-Resistant Pedagogical Framework for Computer Science Education in the Era of Generative AI
Dr. Aalok Thakkar (Department of Computer Science) is pioneering a transformative approach to computer science education that addresses the fundamental challenges posed by the proliferation of foundational AI tools such as ChatGPT, DeepSeek, and GitHub Copilot in academic settings.
The research recognises that traditional introductory programming curricula, which emphasise syntax mastery and simple algorithm implementation, are becoming increasingly obsolete as AI tools demonstrate remarkable proficiency in solving 60-80% of typical introductory assignments. This paradigm shift necessitates a comprehensive redesign of pedagogical approaches to prevent superficial learning patterns and maintain educational rigor.
The research framework employs four interconnected interventions designed to cultivate AI-resistant competencies while leveraging generative AI as a pedagogical enhancement tool. The first intervention involves a fundamental curricular shift toward formal methods and systems thinking, emphasizing proof-based verification, memory management, and program correctness analysis. This approach integrates Why3, a Z3-based verification tool, to provide automated feedback on formal proofs while addressing the limitations of existing auto-grading platforms.
The second intervention develops a pedagogically constrained LLM assistant that enforces structured interaction patterns to preserve cognitive engagement.
The third innovation introduces OOPS (Obfuscated OCaml for Pedagogical Subversion), a custom dialect designed to disrupt copy-paste workflows while encouraging deeper conceptual understanding. This approach employs carefully crafted syntactic constraints including strictly enforced types, pattern matching requirements, and explicit loop invariants to promote active problem-solving without creating pedagogical frustration.
The fourth intervention addresses accessibility gaps through Universal Design for Learning implementation, developing screen-reader optimized tooling and multimodal content delivery systems specifically tailored for the Indian educational context where STEM education remains largely exclusionary for students with disabilities.
This comprehensive pilot deployment at Ashoka University will systematically evaluate the efficacy of these interventions in maintaining educational rigor while preparing students for an AI-integrated computational landscape.
