Gen AI Strategies You Can Use Today: A 4-Part Series

In their new four-part Inspiring Minds article series, Dan Levy and Angela Pérez Albertos of the Harvard Kennedy School share their tested techniques for using gen AI to prepare new courses, improve existing courses, conduct class sessions, and design assessments.

Each article, includes easy-to-follow steps and supporting videos. The theme here is simplicity—so you can jump right into the new term and free yourself up to focus on what matters most: the learning.

4 key principles for generative AI use

There are four foundational principles that anchor how we use gen AI:

• Use AI to assist you—not replace you. Gen AI makes mistakes and has limitations. It will not be better than you at most tasks. But it can help you complete tasks faster or at a higher quality than you’d be able to without it, making you a more efficient, effective educator. So, while it can’t do your teaching for you, it can serve as your copilot.
• Treat AI as a conversation partner. Often, people will be disappointed with the first response gen AI produces and abandon it altogether. But you can and should have conversations with AI. You can give it additional information or instructions to improve its output, just as you might if you were collaborating with a person.
• Remember that you are the expert. You are the expert not only on the subject you teach, but on your students and what will work for them. So don’t feel like you should use what gen AI produces unquestioningly. Evaluate its outputs and refine them according to your knowledge.
• Practice often. The more you use gen AI, the more you’ll understand its strengths and limitations. Practice will naturally make you better at prompting while also helping you discover new ways you can use AI tools to teach.

4 guidelines for great prompting

• Use the TIC formula. This formula ensures that every AI prompt mentions what you want the AI to do (the task), how you want the AI to do it (the instructions), and what the AI needs to know to do it properly (the context).
• When in doubt, ask the AI what information it needs. You can ask it what information it needs, just as you would when developing new class activities.
• When you believe gen AI made a mistake, tell it so.  There’s nothing wrong with pushing back at it, telling it that it didn’t deliver what you wanted, and asking it to try again.
• When asking for a complex task, break it down into smaller components. Providing gen AI with step-by-step instructions can help when asking it to complete complex tasks—just as it does for humans.