2025 Spring · New-TA Peer Sharing (Online)
Opening warm-up/free exchange, mainly Q&A interaction. Panelists are experienced TAs from different courses: Chen Jiajie, Chen Shengqi, Liu Runda, Zhang Yuxuan; they also introduce the rest of the semester's TA seminars and provide recordings for makeup study.
Topics
- Impact of LLMs on TA work: observing students' typical LLM uses (pasting problems to a chatbot, VSCode-integrated LLMs, the model auto-explaining when selecting code during tutoring). Whether and how to allow LLMs is debated without a settled answer. Students face time pressure, but how to build ability? In courses with many written assignments (e.g. computer-systems intro), students ask LLMs first, and some submit LLM output verbatim (about 85% by result) but can't do it in exams. In the last year or two LLM reasoning has improved, affecting assignments more and solving more problems.
- Why teaching debugging is hard: debugging is a tool that takes time to learn and use, with payoff later; students struggle to see its necessity. LLMs now help "look up commands" and even give rough insight/direction like a crystal ball, useful for lower-level courses with easy assignments; but deep, problem-specific debugging still needs students to think line by line, which LLMs can't replace. This is a positive change (lowering the cost of learning to debug). Also explored whether to build "debugging-methodology" prompts so the model does the methodology rather than details.
- Small tips to make TA work more enjoyable: a sense of achievement; the ultimate question is "why are you a TA"; various automation tools (many students); more "learning alongside students" than "teaching students".
- Most time-consuming TA tasks: written-assignment design/grading, writing solutions (which can be enjoyable); one option is to grade a third of students each time; but the longest overall is designing new assignments/labs (may take months); tutoring is easy to overlook but adds up over the semester; grading exams (concentrated in exam week); participating in courseware production, etc.
- Most enjoyable/grinding things: the grinding ones are mainly plagiarism-related, bad attitudes (full slacking), and recording wrong grades.
(Attachment: parts of the 2024 Spring "Efficiency tips and common traps for TA work" for reference.)