Break the Nephron: Designing for Real AI Use, Without the Shortcut
The Assignment
In Anatomy & Physiology II for allied health students (nursing, radiology, dental hygiene, nuclear medicine), student groups invent a fictional kidney disorder and trace how it disrupts nephron function. They pick one mechanism (a transporter, a gradient, a channel, a hormone receptor), deliberately break it, and follow the cascade: what happens to the filtrate, the urine, the blood chemistry, and how the body tries to compensate. The deliverable is a mechanism map plus a logged record of their reasoning.
How AI Fits
Students use AI regularly and intentionally in this assignment, and the concept maps they produce incorporate AI output. The structure is what makes that productive rather than a shortcut. First they commit: mechanism and predictions written down before any tool is opened. Then they use AI as a thinking partner for refining predictions, surfacing gaps, stress-testing compensatory responses, and generating and iterating the map itself. Throughout, they log what the AI suggested, whether they accepted or rejected it, and the physiological reasoning behind each call.
Why It Resists the Shortcut
Because the disorder is invented, there is no correct answer to retrieve and paste. The AI can generate plausible-looking physiology, but it cannot supply the underlying mechanism or judge whether a given consequence actually follows. The student has to drive that, and the log captures whether they did. The result is a spectrum I can see plainly across submissions: at one end, a polished AI map that is confidently wrong; at the other, a student who argued with the tool, corrected it, and iterated toward something sound. The difference is the student, not the tool. The grading rewards mechanistic reasoning over polish: an imperfect map with solid causal logic outscores a beautiful diagram with vague arrows.
Lessons Learned
My early data suggests improvement on related learning outcomes and test scores. On an exam with identical questions across two years, the kidney subscore rose while an unrelated section (respiratory physiology) did not rise to the same degree. However, the sample is too small to claim significance, and I am running a more rigorous comparison next year. In addition, students enthusiastically participated and displayed impressive mastery of a complicated learning objective (kidney physiology) as evidenced by high quality projects and and in-class presentations.