By Haley Martin Sherman
Tired of reading essays that sound strangely identical? Overwhelmed by the sudden influx of conjunctive adverbs like “moreover” and “thus”? Do you feel a sense of nostalgia when you remember silly student mistakes like “doggy dog world”? Does your browser window permanently have three different AI checkers open simultaneously?
…Do you actually have a favorite AI checker? If so, you’re not alone (mine’s GPTZero).
We are living in the AI grading era. It’s frustrating. It’s maddening. It’s the new leading cause of death for instructors everywhere.
But boy, do I have a solution for you!
Use Bloom’s Taxonomy to channel all that frustrated energy into something productive — and alter your assessments.
Remembering (Now With Backup)
Bloom’s Taxonomy has always distinguished between remembering, understanding, applying, analyzing, evaluating and creating. That hierarchy hasn’t changed. What has changed is that the lowest levels — remembering and summarizing — can now be performed instantly by machines that are not going anywhere.
So, if AI can retrieve and summarize information effortlessly, what exactly are we measuring when we grade assignments built primarily on recall?
When I think of “studying,” I conjure up the ancient images of paper-bag-covered textbooks and color-coded flashcards. Historically, memorization mattered. Information wasn’t always accessible, so if you could recall something, it meant you had done the reading, taken the notes and studied. Recall was evidence of preparation.
But in the AI era, definitions are searchable. Timelines are generated. Theories are summarized in seconds — and so are texts. When remembering and summarizing can be automated, assignments that rely on those skills no longer demonstrate preparation, learning or rigor.
This is not an argument against foundational knowledge. Students still need conceptual grounding. But we may want to reconsider how heavily we weight recall in our grading systems.
Technology has caused a shift in culture. It’s time for education to catch up.
Take a look at your assignments. What are you asking students to do? What can be done in seconds by AI? Because if it can be done, you know it is being done.
Then look at your rubric. What categories include formatting, defining, summarizing or even spelling? How much of their grade is rewarded to a robot?
I hate to ask, but…
How much longer are we going to grade grammar?
Where the Real Work Happens
As we move up Bloom’s pyramid, the work changes.
Applying requires using ideas in new contexts. Analyzing requires breaking arguments apart. Evaluating requires taking a position and defending it. Creating requires synthesizing something original.
These levels demand judgment. They demand decision-making. They demand intellectual risk.
AI can generate something that looks like analysis. But it’s rarely original, and it’s rarely deep. For now, robots can’t wrestle (with ideas). They can’t sense when something “feels off,” because, well, they can’t actually feel.
If we deepen the thinking required in our assignments — if we move higher up Bloom’s pyramid — not only do we improve student learning, we have a better chance of AI-proofing our assessments.
If our assignments remain anchored in remembering and understanding, we risk overvaluing what can now be automated and undervaluing the cognitive labor that cannot.
And I’m not having fun reading regurgitated summaries of my favorite texts. I don’t like it. So I’m not assigning summaries anymore.
I try to make positive assumptions. I will assume that my students have done the reading. But, the “reading” they are doing is likely a ChatGPT summary.
So with that assumption in mind, let’s challenge them to go deeper into the ideas, the text, the concepts.
Host a Socratic seminar. Send them out to interview a real person. Have them build something, create something, make something.
Work will be more meaningful, and maybe you can permanently close those three AI-checker tabs in your browser.
For more information, please contact Haley Martin Sherman by email or on Teams.