When AI Can Write the Paper, Make the Reading Matter

As higher education continues to wrestle with generative AI, much of the conversation has focused on what happens at the end of an assignment.

Was this essay written by the student? Did they use AI? Can an instructor detect it?

Professor Jonathan Rees, Professor of History at Colorado State University Pueblo, is approaching the problem from a different direction.

Instead of reducing writing or relying on AI detection, he is putting more emphasis on what students do before they write: reading original sources, selecting evidence, developing interpretations, and defending their own ideas.

A recent faculty case study from the Schlager Digital Library, “AI-Proofing” the History Classroom by Doubling Down on Primary Sources, explores how Rees is designing his history courses around primary source analysis and evidence based writing.

Read the full case study →

And in his upper level American Constitutional Law course, Hypothesis is part of that process.

Primary Sources Make Students Do the Thinking

Rees’s approach starts with the source material itself.

Rather than asking students to work primarily from textbook interpretations, he uses curated primary sources that require students to encounter historical evidence directly. Students must identify relevant passages, integrate quotations, compare documents, and construct arguments from the material they have actually read.

As Rees explains in the case study:

“If you are dependent on AI to do your critical thinking, you’re just going to get a lot of really bad writing.”

Jonathan Rees, Professor of History, Colorado State University Pueblo

The goal is not to create an assignment that technology can never touch. It is to design an assignment where generic answers are not enough.

When students need to work with specific documents, cite particular evidence, and demonstrate how that evidence supports their interpretation, the process of reading becomes much harder to skip.

Bringing That Process Into Canvas

That approach becomes particularly interesting in Rees’s upper level American Constitutional Law course.

Students work with complete Supreme Court opinions from the Essential Primary Sources collection. Rather than reading those complex texts independently and waiting until class to discuss them, Rees integrates the case opinions directly into Canvas and uses Hypothesis for social annotation.

With Hypothesis in Canvas, students can engage with a reading directly inside their existing course environment. They can identify important passages, ask questions, respond to classmates, and begin working through difficult material before the classroom discussion begins.

Learn more about using Hypothesis with Canvas →

The reading itself becomes part of the learning process rather than simply preparation for what comes next.

Make the Learning Process Visible

This matters even more in the age of AI.

A final essay can show an instructor what a student submitted. It reveals much less about how that student arrived there. Social annotation creates another layer of visibility.

Instead of encountering student thinking only when the final paper arrives, instructors can see which passages students selected, what questions emerged, how interpretations developed, where students struggled, and how classmates responded to one another.

This visibility gives instructors an opportunity to engage with student thinking while it is developing.

That does not make an assignment “AI proof” in a literal sense. It does, however, create a learning environment that places greater value on the process AI can otherwise make easy to bypass.

AI Resilient Course Design Starts Before the Final Assignment

Rees’s approach offers a useful lesson for instructors across disciplines.

Responding to AI does not have to mean eliminating writing, returning every assessment to pen and paper, or trying to detect every instance of AI use.

It can mean designing courses around the parts of learning we still want students to practice themselves: reading carefully, questioning evidence, comparing perspectives, discussing ideas, and constructing arguments.

Primary sources create the opportunity. Social annotation makes that work visible and collaborative. LMS integration brings it directly into the course environment students already use.

For instructors thinking more broadly about how to develop these skills alongside emerging AI tools, the Hypothesis AI Literacy Course provides resources for bringing AI literacy, critical thinking, and collaborative learning into the classroom.

Explore the AI Literacy Course →

From Reading to Evidence to Argument

One of the most useful parts of Rees’s approach is that reading is not treated as an isolated requirement.

It becomes the foundation for what students do next.

Students encounter original material, identify evidence, discuss interpretations, and eventually use that evidence to construct their own arguments. In Rees’s courses, those skills are scaffolded as students move from analyzing bounded primary source collections toward broader research and more independent writing.

Social annotation can support that progression by creating a space where the early stages of interpretation are visible.

A student can identify a significant passage. A classmate can offer another interpretation. Someone else can connect it to another source. The instructor can identify a misconception or push the conversation further.

By the time students begin writing, they have already spent time doing the intellectual work that the final assignment is intended to represent.

READ THE FULL FACULTY CASE STUDY

Professor Rees’s complete approach includes primary source selection, scaffolded assignments, evidence based writing, social annotation, and verbal accountability.

“AI-Proofing” the History Classroom by Doubling Down on Primary Sources →

RELATED RESOURCES

• Essential Primary Sources collection, Schlager Digital Library
• Getting Started with Hypothesis in Canvas
• Hypothesis AI Literacy Course