Learning to Get Back Up: Rethinking Assessment in Introductory Physics

A reader considering assessment redesign gets a candid, step-by-step account of converting a large first-year physics course from points-based grading to standards-based grading, including how the weekly quiz reattempt sessions worked, what broke (end-of-semester reattempt pileups, time-consuming explanation gatekeeping, quizzes weighted too low to matter), and how the instructor fixed each problem in later iterations. They will also find one concrete and honestly reported use of generative AI: supplying an original problem and its standard to draft roughly 300 alternate quiz-problem versions, which accelerated the work but still required human review to keep variants aligned with the standard and not too similar to the original. First-semester outcome data (the course failure rate halved from 39% to 19%, average quiz and course grades up around 9-10 points, final exam averages unchanged) gives the reader evidence to weigh against their own context.

Task

Grading & feedback

Time

15 minutes

Level

intermediate

Cost

Free — Nothing to pay. A free account at most.

Tools

None in particular

Published

2026-09-28

Voice

Robert Talbert — Candid in-progress documentation of how AI is forcing a mathematician to redesign grading and assessment.