Recently, I decided to do something old school. I went almost entirely back to paper. Why? Teenagers are CHEATERS. And not in the way that we’re used to. They are no longer copying answers from friends. Or copying & pasting from websites. They’re using AI to write all their assignments. Even assignments that ask for their opinions. 

It’s exhausting and, quite frankly, frightening. It feels like we’re about to have a generation of people who are illiterate cogs in a machine designed to keep us struggling. (No, I’m not being facetious).

It’s even gotten to the point where when I do give a digital assignment, I write things like: “Note: I’m only giving this to you digitally to ensure that you have enough space to write. I expect that you will use your own thoughts and write in your own words. I have absolutely no interest in reading AI-slop. If I suspect that you don’t turn in your own work (I have a good idea of how you write), I will not spend my time grading your assignment.”

This is where social studies teachers are in 2026. According to a March 2026 RAND Corporation report, the percentage of students using AI to complete homework rose from 48% to 62% between May and December of 2025. That is not a slow trend. That is a rapid shift in how your students are completing the work you assign. (I’ve resorted to writing chapter questions that are specifically tailored to the textbook. No more general questions.)

And here’s the kicker: And here’s the kicker: 67% of students say that using AI for schoolwork is harming their critical thinking skills. They know it is happening. They are doing it anyway, in part because we keep assigning work that AI can easily complete.

The solution is not better detection software. The solution is better assignments. Here are five types that AI cannot do for your students.

Type 1: In-Class Discussion-Based Assessment

These are my favorite types of assignments. I have at least one discussion-based assessment per unit in my AP Government & Politics class. AI cannot raise a student’s hand. It can’t make eye contact, respond to a follow-up question in the moment, or respond or create an argument on the spot. Structured in-class discussions are AI-resistant because students must be physically present. You can also restrict their computer use.

The key is making the discussion assessable. Give students a clear discussion role, a specific claim to defend, and a detailed rubric. Also, require students to cite their sources throughout the discussion. This makes it harder for students to use AI.

Fishbowl discussions, Structured Academic Controversy, and Socratic Seminars all work well here. My Culture Wars Fishbowl helps students improve their critical thinking skills. It’s also AI-resistant.

Related: Make Social Studies Discussions Work for Even the Most Reluctant Students

Type 2: Hyper-Local or Hyper-Personal Connection Assignments

Sure, AI can write a decent essay about the causes of World War I. But, it can’t write about your student’s grandfather’s immigration story. Nor can it write one about what the local city council is actually debating right now. The more specific and personal the assignment, the less useful AI becomes.

Try asking students to connect a historical concept to something within a mile of their school. They could interview a family member about a current event. Or they could analyze a local newspaper source rather than a national one. These assignments push students to do something AI cannot do: go out into their actual lives and find the connection themselves.

This kind of local and personal inquiry is also among the most powerful forms of civic education. Research from civic learning programs shows that students who connect historical content to their own communities develop a deeper understanding and stronger long-term retention.

Type 3: Process-Documented Research

AI can produce a polished final product. It cannot show you a student’s thinking as it develops. Process-based assignments close that gap.

Instead of asking only for a finished essay, ask students to submit annotated sources with their own handwritten margin notes. I recently did this with my Civil Rights Pinwheel Activity. I wanted students to write a reflection. But since I didn’t know how much they’d write, I posted a digital copy to my Google Classroom page. Part of the grade was their research. They needed to submit their handwritten research and notes from the discussion to receive full credit. Other ideas include:

• a revised outline showing how their argument changed
• a voice memo explaining why they chose their thesis
• a research journal tracking their questions over time.

These tasks are harder to fake because students have to put in the thinking themselves. They also help students have a greater depth of knowledge on the topics.

The CITE Journal’s research on AI integration in social studies education recommends framing AI as a tool that students engage with critically and transparently, rather than banning it outright. There are helpful and ethical ways to use AI. It’s important that we teach students how to use it that way. Process documentation is a powerful way to do that: students can use AI as part of their research process. But they must document how their thinking diverged from the tool’s output.

Type 4: Real-Time Observation and Analysis

Getting students into the action can help AI-proof your classes. AI cannot watch a city council meeting. It cannot observe a courtroom, attend a community forum, or report on what actually happened at the school board meeting last Tuesday. Real-time observation assignments force students to engage with the world as it happens.

Assign students to watch a government body in action, whether in person or via a public livestream, and complete an assignment based on it. Ask them to identify which constitutional principles they see and what questions they have.

According to Fortune’s investigation into the reasoning crisis in schools, frequent, unguided use of AI is associated with lower critical-thinking skills among young people due to increased cognitive offloading (the use of external tools to reduce mental load). Observation is the antidote: it requires students to think in the moment, without a tool to think for them.

Type 5: Student-Created Teaching Artifacts

One of the clearest signs that a student understands something is their ability to explain it to someone else. Ask students to create something that teaches a concept to a specific audience. These could include

• a short video for incoming freshmen
• an explainer for parents
• a visual guide for middle schoolers
• a presentation to the principal about a civic issue in the school.

These assignments require students to know the content well enough to create a meaningful product. One that will teach someone something. AI can generate generic explainer content. It cannot produce something that actually works for a specific audience that the student knows personally.

Research on AI-resistant assessment design published in a 2024 Frontiers in Education study found that assessments requiring creativity, audience awareness, and personal application are the hardest for AI to replicate. Student-created teaching artifacts check all three boxes.

Better Assignments Win

AI is not going away. And the temptation for students to use it to cut corners is not going away either. But you should not have to choose between going back to paper and watching students submit AI slop. (Although I’m very happy that I went back to paper.)

The five assignment types in this post have something in common. They all require students to show up. In person, in their communities, in their own thinking. They ask students to do things that a chatbot simply cannot do for them. And when students actually have to engage with the material, something shifts. They start to think. They start to care. That is the whole point.

You went into teaching because you believed students were capable of real thinking. Do not let AI lower the bar. Raise the assignment. The students who learn to think clearly and independently in your classroom are the ones who will be ready for whatever comes next.