I want to revisit a prior post to linger on the three design moves for faculty because this is where the AI conversation becomes practical.
If AI changes the conditions under which students produce work, then our assignments need to do more than produce finished products. They need to help students practice judgment.
First, make process visible.
This does not mean adding busywork or turning every assignment into a surveillance exercise. It means designing moments where students show how they are thinking: an early draft, an annotated source, a revision note, a design rationale, a short reflection on what changed and why. When process is visible, faculty can respond to learning while it is still forming. Students also become more aware of their own choices, which is the beginning of agency.
Second, build comparative iteration.
In an AI-mediated environment, the first answer is easy to generate. That does not make it appropriate. Students need practice comparing possibilities. Which explanation is stronger. Which source is more credible. Which version sounds more precise. Which approach fits the audience, purpose, and discipline. Comparison slows the rush to accept the first fluent answer and turns attention back toward judgment.
Third, protect voice and judgment.
Students need opportunities to make claims they can stand behind. That means assignments should invite specificity, context, and consequence. Ask students what they chose to include, what they left out, what evidence mattered, what tradeoffs they accepted, and why. Ask them to connect their work to a local problem, a field-specific question, a lived observation, or a decision that requires accountability. Voice is not just style. It is the sound of a person taking responsibility for meaning.
These moves are not anti-AI. They are pro-learning.
They help faculty design assignments that do not depend on pretending AI does not exist. They also help students develop the literacies they need to use tools without disappearing into them. The goal is not to make every assignment AI-proof. The goal is to make thinking harder to bypass and easier to support.