By this, I mean: as soon as we define a winning condition, we enable an assessment task to be played like a game. Any task that involves assigning a quantitative value, even if that value is 0/1 (fail/pass), can be gamed. Assigning value like this requires rules, and rules invite gamesmanship.
For example, if a task is graded via a rubric, we have to match the student's work to specific descriptors on the rubric. If a student wanted to game this using GenAI, they might feed the rubric descriptors into an LLM. They could also pass them to a contract cheating vendor. Or, following the maxim “Ps get degrees”, a student can simply aim at the lowest-level descriptors to exert the minimum effort required to pass the assessment. And the requirements for a “pass” mark are often well below a standard we might actually consider passable.
So we need to be very aware of what kinds of games we make possible to play — and when and why we would consider such games “cheating”.
If students should be able to perform a task to a very specific level of precision (say, applying a mathematical equation or performing heart surgery), then a rubric-based assessment strategy is not appropriate. “Ps get degrees” should not apply here. I would argue that a medical student who “completes operation with minor inaccuracies” is not ready to practice surgery in the world.
Banning generative AI is also inappropriate in this case, because such tasks should be provided optimal conditions for perfect completion, not artificial barriers. Isn’t the point that the task is done correctly and safely? If the task is best done correctly and safely with particular digital (or other) tools, then the presence of those tools should validate, not invalidate, the assessment evidence.
Conversely, it may not be appropriate to use grading at all. If an assessment is for the purpose of diagnosis (the teacher identifying the student’s needs or work style), then a descriptive approach involving analysis and feedback may be the way to go.
It’s pretty difficult to “game” an assessment that has no winning conditions. All a student has to do is submit something. If it’s something pointless, then the feedback they receive will also be pointless, and they have wasted the opportunity to learn from it.
Obviously, there are times when we do need to produce a quantitative judgement. Most importantly, we need to know when:
a student is not ready to progress to further learning
a student is ready to practice a real skill with real risks in the real world.
To my mind, these are the two reasons why we would need to judge student work in a quantitative way. So, in the next proposal, I offer a way of thinking differently about teacher judgements of readiness for progression.
References
Blum, S. (2020). Ungrading: Why Rating Students Undermines Learning (and What to do Instead). West Virginia University Press. https://doi.org/10.31468/dwr.881
Dawson, P. (2025, February). Grade inflation, marking to a curve, or standards creep [Post]. LinkedIn. https://www.linkedin.com/feed/update/urn:li:activity:7300303779692756992/
McTighe, J., Brookhart, S. M., & Guskey, T. R. (2024). The Value of Descriptive, Multi-Level Rubrics. ASCD. https://www.ascd.org/el/articles/the-value-of-descriptive-multi-level-rubrics
Reynoldson, M. (2025, July). Where to from here? Foster educational relationships. The Mind File. https://miriamreynoldson.com/2025/07/01/where-to-from-here-relationships/
Reynoldson, M. (2025, March). Soaring standards in the age of AI assessment.The Mind File. https://themindfile.substack.com/p/soaring-standards-in-the-age-of-ai
No posts
Comments
Nothing yet. Say the first thing.
Sign in to join the conversation.