An instructor guide to easing into generative AI

We understand the doubts and misgivings that most instructors have about generative AI. Learning new software and revamping assignments can be time-consuming. And in many ways, it feels as if control over classes has been stripped from professors, with AI threatening the integrity of assignments and assessments even though proven alternatives for developing writing, coding, problem-solving, critical thinking, and other high-level skills have yet to emerge.

We don’t have all the answers about generative AI. No one does. That’s what makes its use exciting (even as it produces anxieties). Don’t expect students to know how to use generative AI effectively either. You do want to bring them into the learning process, though, and to have them share their experiences with learning new processes and exploring a new type of technology. Rather than looking at AI as an encroachment, look at it as a new tool with new possibilities.

So set aside your doubts, at least for a while, and consider a few steps you can take to learn about generative AI and to use it effectively in your classes.

Get to know how generative AI works

This is an important first step. You don’t have to become an expert at every platform or understand the intricacies of algorithms or the large language models that make generative AI possible. You should understand basic prompting, though. (Prompting involves asking questions of and providing direction to chatbots so that they return the types of information, images, code, or calculations you want.) You should also get a sense of the capabilities and the outputs of generative AI.

We have created a separate guide on prompting chatbots and tutorials on getting started with the four most popular models: ChatGPT, Microsoft Copilot, Gemini from Google, and Claude.

If you have never used AI, we suggest starting with Microsoft Copilot. It uses the same platform as ChatGPT but adds a search component. It is free, and if you use your KU account to access it, you will have additional privacy protection. 

Learn to create good prompts

There are no secrets. You use natural language (complete sentences) to direct the chatbot on what you want it to do. Assign it a role, provide context for your query, and provide specific guidance on what you want it to do and what form you want the output to take. For example, you might try something like this:

Act as an expert in generative artificial intelligence and higher education. Create a two-paragraph policy on use of generative AI in a 100-level economics class. Include bullet points stating that if students use generative AI, they must include a reflection statement how they used AI, how they incorporated AI material into their own work, and what they learned from the process of using generative AI. Include a warning that material that is simply copied and pasted from a chatbot will be considered academic misconduct.

You can then ask follow-up questions or ask the chatbot to elaborate on or explain any aspect of what it has produced. For a deeper dive, we recommend reading our prompting guide, which includes examples of how others approach prompting.

Guide students in idea generation

This is an easy first step in bringing generative AI into your classes. Nearly every class has some assignment that requires ideation, and chatbots can help students with that process. They can also provide feedback on narrowing the overly broad topics that students usually start with. Some can even point students toward appropriate literature. A separate page shows in detail how a generative AI chatbot can guide a student in a research paper. Here’s an idea of how an assignment might start. You would need to provide specifics about your class, discipline, and expectations:

The first step in any research project is an idea or observation. Use the chatbot of your choice to explore ideas for your project. You can ask it to provide ideas about researching certain aspects of (your field) or you can start with your own idea or observation and have it lead you through a process of refinement. Once you are satisfied with your topic or research question, put it at the top of a page and include all the interactions you had with the chatbot (including your prompts) after that.   

Outline a project

This is another area where generative AI can be especially helpful in guiding students with a project. Once students have an approved topic or research question, have them use a chatbot to create an outline or suggest the steps they will need to take to complete the project. This sets up a good format for an iterative process of developing a project because you can ask the chatbot to elaborate on any or all of the steps. Here’s how you might ask students to approach the assignment.

Now that you have an approved topic for your project, I would like you to use a chatbot of your choice to create an outline of major areas you will cover and the steps you will take to successfully complete your project. Start with the prompt below. After that, it will be up to you to ask follow-up questions so that you understand the approach, the terminology, and the steps involved. Keep in mind that chatbots often get things wrong. It will be up to you to evaluate the information the bot provides and to revise the outline and steps as needed. You are in charge of the assignment. The bot isn’t. When you are done, turn in your outline and the steps you plan to take with the project. Include a transcript of your interactions with the chatbot, provide a critique of the bot’s output, and explain how you arrived at the outline you have turned in.

Opening prompt: You are an expert in (your field). You have been assigned to complete a project on (student’s project, questions and any necessary context). Create an outline with six major areas to cover and provide five steps you need to take to complete the project.

Summarize research

Most generative AI platforms do a relatively good job of summarizing papers, reports, websites, and other material. By tapping into that ability, you can help students learn to use generative AI to determine whether a paper or other information fits with their interests, and to improve their ability to draw in more sources from a broader range of specialty areas.

There are several ways you might approach a summarization assignment. For instance, you might have students use different AI models to summarize a complex paper and then ask them to compare and critique the results. Which was most accurate? Did any of them miss key points? If so, what? How might they phrase AI prompts differently to get better results?

Another approach is to have students create their own summary of the article in class without the help of generative AI. Afterward, have them use AI to create summaries. Then ask them to compare and critique the differences. How did their work compare with the output of the chatbot’s? Did the chatbot provide information they wish they had provided in their summary, or did the chatbot fail to pick up some key points? Also have students consider how different prompts might result in different results of summarization.

Simplify concepts

Nearly every field has concepts, language, writing, or other material that leaves students baffled. Encourage students to use generative AI to explain that material in ways that help them understand. For example, you might have them take a complex research paper and ask a chatbot to explain the concepts and reasoning in the paper for a general audience. Or you could have students experiment and have AI create analogies that make a particularly challenging concept more accessible. This could be done in class or as homework assignment, with students sharing examples and explaining the prompts they used to get the shared result.

Critique drafts, outlines and sources

Students need good feedback on any type of work if they are to improve. Many classes use peer review in addition to instructor feedback to help students improve their work. Generative AI can add yet another round of feedback. This could be over a draft, a list of sources, or any other part of a project. It could address the quality of the writing or the code or the strength of an argument. Students could have generative AI critique an outline or a plan for completing a project. Are the steps appropriate? Are things missing?

All of these approaches will require experimenting with prompts. For instance, an AI chatbot will need to know such things as discipline, audience, and style to provide feedback on writing. It will need to know how code will be used, whether the code is part of a larger program, and what operations the code is intended to inform. In many cases, generative AI will make appropriate inferences, but it is better to provide clear direction at the start.

One approach to this process would to have students seek feedback from a chatbot at the same time a colleague or the instructor is completing a critique. Then students can compare the advice they get from the various sources. How did the advice differ? In which areas did the individual sources provide the best feedback? What did the student learn from the exercise that might help them create better generative AI prompts next time?

Explore and evaluate different AI tools

The number of generative AI tools and sites is enormous, and many new tools appear weekly. By having students explore these tools, you can help them find ones that will be most useful to them and flag ones that aren’t worth the effort. Provide a list of tools in a spreadsheet and have students sign up for the tool they plan to explore. (We include a list of potentially useful tools on another page.) If they know of a tool not on the list, they are free to add it.

Once each student has chosen a tool (or several if you prefer), they should experiment with it and write a review. What is it and how is it intended to be used? Who is it for? What are its strengths and limitations? How difficult is it to learn and use? What other tools does it compete with or what others have similar capabilities? Are there potential uses beyond those the tool’s creator lists? Is it something appropriate for your discipline?

Once all students turn in their reviews, you can aggregate them and provide a working list for everyone in the class. That will expand students’ understanding of generative AI tools and will help guide you on tools to consider in future iterations of your course.

Critique AI-created materials

Critical examination of materials created by generative AI is an important skill for students to build. One way to approach that is by having a chatbot create an example of an assignment students will do themselves. In groups, have students talk about the strengths and weaknesses of that material, whether it be writing, coding, calculations, design, images or something else. What are the strengths and weaknesses of the work? What signs can students see that something was created by generative AI?

Relatedly, you can give students an AI-created work and have them use AI to critique it. That gives students practice in prompting and in seeing how chatbots offer advice. How accurate is that advice? How useful? Does it go beyond the mundane ,or does it provide concrete actions that could be taken to improve the work? This exercise has the added benefit of helping students build their own skills at providing feedback.   

Provide personalized advice for students

Generative AI can create different versions of readings, assignments, and approaches, tailoring them to a student’s needs. So if a student is struggling with a concept, you can have the student use generative AI as a tutor and lead the student in a review session. This might involve having the chatbot create questions for the student to answer or problems to solve to gauge where the student might be having difficulty. Then you can prompt the bot to provide feedback on that work and additional follow-up work.

This approach also works with things like career exploration, study tips, and anything else a student might want to explore.

A caveat: Generative AI will make things up at times or get things wrong. In experimenting with problem-solving, we found that generative AI performed the math incorrectly and gave a wrong answer for an example it had created.

Reformat references

Generative AI can take references created in, say, Chicago style and reformat them in APA or MLA style. This can be a big time-saver for complex articles. It can also be a great learning opportunity. Having students work with generative AI in this way can help them learn different types of academic style and help them train their eye and habits on a style for their discipline. It can also lead to discussions about why different disciplines follow different styles.

Create rubrics

The first versions of rubrics can be challenging to create. As with any other type of writing, generative AI can jump-start the process. Instructors have had varying degrees of success with having AI do this. The goal isn’t to create a perfect rubric on the first attempt, though. Rather, it is to provide a starting point. A rubric requires clear learning goals and a clear understanding of how those goals will be measured. Working through the creation process with generative AI can help instructors gain valuable feedback on their course goals, outcomes, and rubrics. It can also help them improve their prompting skills, especially with follow-up prompts that will be needed to shape a rubric.

Add a quasi-methods section or a reflection section to papers

Use of generative AI is a new method for many aspects of academic and non-academic work. Having students create a short methods section on assignments or explain their use of AI in an existing methods section is a good way to hold them accountable. What did generative AI contribute to the assignment? How did students adapt the AI output? For instance, did they have AI create a draft that they adapted and added sources, depth, and direction to? Did they ask AI to do an outline? Did they have it provide feedback on their work?

Having students address those sorts of questions can help them understand the importance of transparency. By thinking of generative AI tools as part of a method, we can also help students learn more about their own strength and weaknesses. None of us is good at everything, and learning how generative AI can augment our personal skills is important to know. Encouraging students to reflect also helps them build metacognitive skills – skills that help them become better self-learners.

Create learning games for class

Generative AI has a playful side, and instructors can tap into that to gamify aspects of class or to create learning games that reinforce important ideas or concepts. Many instructors already use Jeopardy-like games to gauge students’ understanding of course material. Generative AI can easily create those types of questions and answers for nearly any topic. It can also guide you through creating a new game based on the types of questions you want students to answer. There are also AI-enhanced specialty tools you can draw on. Some are free. Others have at least a free tier, with additional functionality available at a cost. Here are some examples:

• Are You Smarter Than ChatGPT?is a free game that gives you two versions of famous quotes, one with the original ending and another generated by ChatGPT. You have to choose which is real. 
• Booom and Conkerare free tools for creating quizzes and games.
• Brevity 500provides examples of poorly written business messages you can edit and check against an AI analysis.
• GPTGameallows you to describe a game you want to play and have ChatGPT create a prototype.
• GPT Games uses a similar approach but also allows you to play games others have created.

Still feeling stuck? Ask generative AI for help.

No doubt this advice will grate against the academic sensibilities of many instructors. Before you dismiss the idea, though, try it. Other professors have been using ChatGPT, Bing Chat, and other generative AI tools to develop class materials.

Andrew Maynard, an Arizona State professor, created a class with the suggestions of ChatGPT. He had ChatGPT act as a tutor, asking students questions about weekly class topics in prompting for generative AI, and grading them on their responses. The class used a competency-based approach, with students allowed to repeat the process until they got the grade they wanted. Students then turned in transcripts of their interactions with the software. In writing about that, Maynard said: “It’s almost as if ChatGPT is fine-tuning my brain to be a better instructor by enabling me to see in intimate detail how students interact with it, and how new approaches to learning can leverage this in ways that lie far beyond conventional approaches to education.” (Maynard wrote about his experiences in Slate.)

You wouldn’t have to go as far as Maynard in using generative AI in a course, but asking generative AI for ideas in approaching your course could help you move beyond a feeling of helplessness that many instructors have expressed. For instance, you might create a prompt like this:

Act as a college professor in (your field) who has limited experience with generative artificial intelligence. You are worried that students will use generative AI to avoid coursework intended to help them gain skills they need to succeed in future classes and in careers. I’d like you to create two things: First, a course policy that balances student use of generative AI with the need for students to gain core skills. Second, an assignment that encourages the use of generative AI but that would require students to apply critical thinking, creativity, and application of disciplinary understanding. For that assignment, create a paragraph of context and a step-by-step guide for students to follow. Also create an instructor guide on how to assess the work.