🧠 AI Ed Wiki

📄 Research Article

Test-Driven, AI-Assisted Learning: Replacing Lectures with Weekly Closed-Book Tests

◐ medium · 2026-06-23cs-education, llm, active-learning, assessment, higher-ed, automated-grading, scaffolding

Liu et al. (2026) report on a 13-week Test-Driven, AI-Assisted (TDAA) redesign of a Theory of Computation course at HKUST (Guangzhou). The course replaced all lectures with self-directed, AI-assisted learning and weekly closed-book tests serving as high-frequency quality gates. AI agents helped the instructor prepare learning paths, course websites, test drafting, grading workflows, and content repairs — all managed through a version-controlled agent workspace. Student survey data (N=18), weekly scores, and git history suggest the model preserved individual accountability while making material production and marking scalable with human oversight. The work contributes a reusable design pattern for AI Changing Teaching Workflows that integrates Active Learning principles with LLM-powered scaffolding, and offers practical implications for Assessment design in Higher Ed contexts. The approach connects to broader conversations about Automated Grading and LLM Student Modeling Memory as enablers of scalable personalized education.

Connected Concepts

  • Active Learning
  • LLM
  • Assessment
  • Higher Ed
  • Automated Grading

    • Connected Articles

  • AI Changing Teaching Workflows
  • LLM Student Modeling Memory

    • Citation

    Jin-Guo Liu, Shang-Qi Lu, Xin-Ran Shi, Long-Li Zheng, Wei Wang (2026). Test-Driven, AI-Assisted Learning: Replacing Lectures with Weekly Closed-Book Tests. arXiv:2606.23315. arXiv:2606.23315 (cs.CY)