Harvard AI Pedagogy Project Remixes

Instead of writing a standard book report, students "interview" a character from their current reading list. This reveals if students actually understand the character's motivations and tests the AI's ability to stay in character.

Students are worried about AI taking their future jobs. This lesson empowers them to use AI to analyze career paths, identify necessary skills, and build a "future-proof" action plan.

Students act as worldbuilders to design a new planet. They use AI to visualize the planet, ensure the physics makes sense, and create the backstory for a sci-fi narrative.

Textbooks can be dry. In this lesson, students drop themselves into a specific historical moment (e.g., The Storming of the Bastille) and use AI to write a sensory-rich, first-person narrative of the event, which they then must fact-check.

Students often argue past each other because they aren't arguing about the same kind of issue. In this lesson, students use AI to analyze their own topics through the ancient Greek framework of "Stasis Theory" to pinpoint exactly where the disagreement lies: Fact, Definition, Quality, or Policy.

"P(doom)" is the probability that AI will cause a catastrophic event. In this lesson, students move beyond sci-fi tropes to research actual existential risks (and benefits), eventually assigning their own "probability score" backed by evidence.

Students explore how easy it is to rewrite history visually. They will create a "Two Truths and a Lie" quiz using real historical photos and AI-generated fakes, challenging their classmates to spot the difference.

Most people listen only to respond, not to understand. In this lesson, students must "program" the AI to follow "Rogers's Rules" of active listening-where you must restate the other person's point to their satisfaction before arguing your own.

Headlines claim AI is "becoming sentient" or "will steal all jobs by Tuesday." This lesson turns students into fact-checkers who analyze news articles to determine if they are legitimate reporting or just clickbait "hype."

How does a computer know that a circle with a stick is a "lollipop"? In this fun, gamified lesson, students help train a neural network by drawing, then analyze how the AI guesses what they are drawing in real-time.

Textbook language is often different from "real" language. Students prepare to interview a heritage speaker (a family member or community member) by first rehearsing the interview with an AI persona, comparing the "robot's" code-switching to the real human's.

Does the "vibe" of an AI change how much we trust it? Students will interrogate the "personality" of a chatbot by comparing a neutral interaction against a specific "persona" to see how tone influences credibility.

Imagine you ask your favorite teacher for a college recommendation letter, and they use AI to write it in 30 seconds. Is that cheating? Students will generate their own letters to debate the ethics of automation in relationships.

If you mix a Picasso painting and a photo of your dog using AI, who owns the result? You? Picasso's estate? The AI company? Students remix images to explore the messy legal world of AI art.

Instead of the teacher dictating the AI rules, the class will debate and draft their own "AI Constitution." Should we ban it, restrict it, or embrace it?

It is dangerously easy to generate fake news. In this "Red Team" exercise, students try to trick the AI into writing a convincing article about a harmless false topic to learn how misinformation spreads.

Can a robot understand the human soul? Students compare a literary critique written by a human expert against one written by Gemini to see what the AI misses (or gets right).

Seeing is no longer believing. Students will choose a historical hoax and use AI to create "photographic evidence" that makes the lie look real.

Can an algorithm feel heartbreak or joy? Students challenge the AI to create original metaphors for complex emotions to see if it can be truly creative or just cliché.

You clicked "I Agree," but do you know what you agreed to? We will translate the "legalese" of AI Terms of Service (TOS) into plain English to see what data we are giving away.

If you ask AI to draw a "CEO," who does it draw? Students will expose the hidden stereotypes in AI models by generating images of people in various professions.

Students play a digital version of the 1920s Surrealist game "Exquisite Corpse." They will co-write a story with AI, taking turns writing paragraphs to see how human creativity blends with algorithmic prediction.

Instead of asking AI to write code for them, students write their own Python game first, then use AI as a "QA Tester" to find bugs and edge cases they missed.

Students flip the script: instead of trying to get the AI to write a good essay, they try to force it to write a bad one (clichés, errors, hallucinations). Then, they play "teacher" and grade/fix it.

A research project structured like a sandwich: AI helps brainstorming (Bread), the student does real human interviews (The Meat), and AI helps outline the results (Bread).

Students have a philosophical debate or dialogue with an AI that is roleplaying a specific persona (e.g., Socrates, a confused alien, a cynical detective) to refine their own thinking.

Can AI actually write like Shakespeare or Steinbeck? Students test AI's ability to mimic famous authors and analyze where it succeeds and where it fails to capture the "soul" of the style.

Students become "Forensic Image Analysts." They look at viral AI images to spot the fakes, then try to create one (if tools allow) to understand how easy it is to manufacture "evidence."

Nervous about an interview? Students use AI as a simulator to practice interviewing a "person" before they have to do it for real.

Students turn the tables and become the grader. They feed a difficult math or science problem to the AI, ask for an explanation, and then determine if the AI is right, wrong, or hallucinating.