Objective: Students should be able to evaluate appropriate use of generative AI in academic and/or professional contexts.

Demonstrates using AI successfully requires critical analysis and multiple stages of revision.

1. Image Generation (with or without use of generative AI)

Option 1: Have students review the final book cover image for SOURCE and the series of images leading up to it. Ask them to reflect on their progression and how the authors describe their iterative process. [I’m still looking for a better example that instructors could use.]

Option 2: Create an image yourself using generative AI and document the revision process. Present the progression of images and your revision process. Ask them to discuss that progression and process.

Option 3: Have students create their own image and document and reflect on their revision process. This exercise could take many forms—from a formal essay to a brief in-class individual or group activity.

2. Sequential Prompt Revision from Clark (requires use of generative AI)

Step 1: In teams, ask students to discuss possible answers to a class-related question. Clark asked students in a Marketing class to come up with “good target markets” for a particular product. Students then ask AI for good target markets and discuss the differences between their ideas and the AI output. Students note their prompts. A whole-class discussion comparing original ideas, new ideas generated, utility of AI use, and prompts follows. Instructors should note how different approaches in prompts elicited different results.

Step 2: Teams choose their preferred result from step 1 and ask AI a related follow-up question. Clark had students ask AI for a “positioning” or key benefits for their chosen “good target market.” Again, the utility of results and difference in prompts are discussed with the whole class.

Step 3: Any number of questions that build on previous ones can follow. Clark had students take the “positioning” generated in step 2 and ask AI to suggest creative product and pricing ideas.

Final step: Ask students if they could have come up with these ideas on their own. Below are some of the discussion prompts from Clark.

What were your assumptions about [topic]? Does AI change them?
What were your assumptions about how generative AI would work? Did this exercise change them?
Why did you use that prompt?
Enter the same prompt again/regenerate the response. How do the outputs change?
Give your AI more context/have it imagine it is a kind of person. How do the outputs change?
How would you decide X was a good idea? What else would you want to know? How would you test it?

B. Critical Evaluation of Context

Encourages thinking about the affordances and limitations of Gen AI use in different contexts.

1. Cake-making analogy (no requirement to use AI) from Bali

From Bali’s abstract: “The cake-making analogy equates different ways of acquiring a cake (baking from scratch, using a readymade mix from a box, buying from a bakery or buying preserved cake from a supermarket) with varying degrees of reliance on AI as a shortcut for tasks or assignments. The lesson invites participants (who may be students or teachers) to critically consider the implications of each mode, examining factors such as quality, time, cost, and personal investment. This analogy is then applied to generative AI, prompting discussions on essential learning outcomes of a course or of a particular assignment, quality of output, learning process, and the ethical considerations of AI use. ”

While there are too many examples and suggestions for discussion prompts to duplicate here, the activity is to discuss when and why the four iterations of cake-making are appropriate, extend that discussion to the use of generative AI, then create guidelines for when it can be used for that particular class (or assignment). The activity may lead to interesting questions about quality of cake/work, resources available to students, when it is ok to use a cake mix/template of sorts, when is it ok to “bring a twinkie” to a gathering?

This line of thinking could fit other contexts that the instructor and students may be familiar with. For instance, a discussion of when it is appropriate for student to write their own code (making a cake from scratch), use existing open-source code (cake from a box), pay for access to a closed-source library (like buying an expensive cake from a bakery), use their team/research group’s code repository, or write code with a generative AI assistant (e.g., GitHub Copilot).

2. Using predictive text as authoring tool (Denial, 2024) (requires at least instructor use of generative AI or access to an AI-generated text)

Option 1: Analyze a paragraph or document generated largely by predictive text.

Option 2: Have students generate their own document using largely predictive text. This could be as simple as having them use their phone in class to generate a paragraph on “the history of yesterday.”

Denial (2024) notes that “ChatGPT and other similar products do not generate knowledge, but instead work by means of sophisticated predictive text operations. . . . I asked students to write the history of yesterday using only predictive text, and we shared the results to hilarious effect. The stories were, predictably, bland, vague, and very absurd, which I linked back to the assigned reading about how LLMs worked. This is what generative AI does, I suggested – it makes educated guesses about which words will come next in any given phrase. This offered a great segue into a discussion of whether generative AI can be useful with and/or without editing, and what judicious editing might look like. (Wieck, 2023.)”

This activity could be extended to answer a prompt relevant to the class.

C. Idea Generation

The below activities are documented in detail in Tsufim & Pomerleau’s (2024).

Option 1 Brainstorming: Have students brainstorm their own ideas regarding a relevant prompt, then immediately ask ChatGPT or another AI tool to brainstorm ideas on the same topic. Compare results and potentially combine them.

This activity could be done in different ways: individually, in groups, as a class.

Option 2 Generative AI “peer review”: Before sharing their work with others, students “run it through” generative AI. Tsufim and Pomerleau (2024) found that “Students unanimously agreed that running their own initial, sometimes vague ideas through generative AI not only gave them more material to work with, but also allowed them to see the strength in their own ideas.”

Some of the discussion questions provided by the authors are below.

Questions to Guide Reflection and Discussion (Tsufim & Pomerleau, 2024)

How does the use of generative AI for brainstorming challenge traditional notions of creativity in the writing process?
Discuss the potential for AI tools to standardize language and ideas in academic writing. How might this affect students from diverse linguistic backgrounds?
Consider the role of anonymity in AI-assisted brainstorming. How does it affect students’ willingness to share and refine ideas?
Consider the balance between efficiency and depth in the brainstorming process when incorporating AI tools. How might this affect the development of critical thinking skills?

D. Critical Reading

Watkins (2024) asserts that “reading assistants like Explainpaper and SciSpace are powered by large language models like OpenAI’s GPT and can help students augment reading. This application of generative technology could aid non-native speakers, students with disabilities, and those struggling with reading comprehension.” He integrated their use into his first-year writing class. In his assignment prompt, he requires students to “upload a PDF of the paper you want to understand better . . . . The AI will read and analyze the full text. Then you can have a conversation with the AI assistant to ask questions about concepts, terms, themes, and how different sections relate to each other and the overall arguments or themes in the paper. The AI reading assistants leverage the power of generative AI to understand your questions and provide clear explanations tailored to the content of the specific paper you uploaded. This makes it easier to grasp difficult material and see connections between ideas when reading complex papers.”

Watkins found that “AI reading assistants like Explainpaper and SciSpace made challenging academic readings more accessible to students. Many students appreciated how the AI apps reworded confusing sections into plain language, making dense readings more accessible.”

Reflection Questions (Watkins, 2024)

How did the AI reading assistant impact your engagement with and motivation to read the essay? Did the technology make the reading experience more enjoyable or interesting? Why or why not?
Consider what was offloaded to the AI and what you had to learn or understand independently. How did this division of labor between human and machine affect your learning process?
Do you feel more inclined to rely on technology for comprehension in the future? What are the potential benefits and drawbacks of this dependency?
How did using the AI reading assistant affect your ability to analyze and critique academic papers? Consider whether the technology helped develop these skills or if it overshadowed your analytical process.
How did the AI app enhance my understanding of the difficult PDF I was reading?
What were some of the key features of the AI app that made it easy for me to read and comprehend the PDF?
Did the AI app provide additional resources or contextual information that helped me better understand the content of the PDF?
Did the AI app help me to focus and stay engaged with the challenging content of the PDF? Why or why not?
Did using the AI app change the way I typically approach reading difficult material?
What was my overall experience using the AI app, and how would I rate its effectiveness?
Did the AI app help me to retain more information from the PDF than I typically would have?
Do I see myself using an AI app in future reading activities?
Did the AI
app help to reduce the stress and anxiety associated with reading difficult material? Why or why not?
Would I recommend an AI app to others who struggle with reading and understanding complex PDFs