An AI Activity to Try With Faculty

Sharing some new activities I've been trying out with faculty

Last week, I did a workshop with Post University. It was actually a follow up to a session I did with them back in April. So, an encore? They were one of the first folks to invite me to come to talk to their faculty in Spring 2023. You can see the slides and resources from that and several others in this post. That even led me to make this video (32:15 minutes) which was a melding of a lot of my talks in the spring.

I’m not going to repost the talk itself because much of it is not particularly new ground to some of my previous posts here (it follows a similar path to this one I shared earlier this month). You can, of course, check out the annotated slide deck if you want to see it anyway.

But I do want to talk about what was new in this workshop that I hadn’t done elsewhere. This time around it was starting to build out some activities for faculty to work through as they considered the institutional policy and what or how they might think about it in terms of working with it.

Now, the timing of things was off and so while I had hoped to do 2 activities, I only got to do one of them. Still, I’ll share the one that we did and my reasoning and lessons learned from it as well as how I constructed them. You can get the full guidance and details in this resource.

DALLE-3 with prompt: “humans and humanoid robots sitting around a table sharing drinks and laughing. Style: 19th century American folk art.”

Activity: Edge Cases with AI Policy

One of the things about policies is that they are nice and useful but they don’t necessarily prepare you for edge cases. And I think with generative AI there will be lots of edge cases. A policy might lead you in a particular direction but that can often only be once you, yourself, have a solid understanding of your own feelings about what the policy is and how you make sense of it in your particular context (i.e. class).

So I wanted the faculty to practice with some edge cases related to the institution’s policy and their particular discipline.

Here’s what I did to prepare for the activity—and yes, it relied heavily on generative AI to do this.

Prompt #1: Processing the Policy

First, I wanted to create edge cases and do so in a way that felt like every faculty member could tap into in specific disciplines. The general scenario would always feel too elusive for the different courses. So I started with the following prompt:

You are an expert in policy development and education.  The following is a generative AI policy for Post University.  Review it and analyze it.  Then create 10 case study descriptions related to teaching and learning in specific course contexts that might challenge or locate loopholes in the policy.  

Do not repeat the same issue and make sure to change courses across disciplines while also the details and dynamics of each case study. 

"[Institutional Policy for Generative AI]"

I did this prompt to both feed ChatGPT the policy and to prime its work as well as to see what kind of outputs it would produce. I did not intend to use the 10 case studies but just to see the flavor of its thinking because I tried something more substantial.

Prompt #2: Refining the Quality of the Case Studies

Now that I saw the list of what kind of case studies it would produce. I wanted to reinforce what I thought were solid responses and use those as future examples. So I provided this prompt.

Examples [identify which #s are good] are good examples.  The rest feel straightforward.  

Create new case studies keeping that in mind and really aim for nuance and challenges that both the faculty and the student might face. 

Remember that the goal is to find things that around using generative AI that aren't covered by the policies and are truly gray areas for consideration. 

Building upon the successful and trying to really clinch the challenging aspects was really important here. I wanted most if not all faculty to feel like there was a case study they could engage with.

Prompt #3: Building case studies for each discipline

Once it produced better examples, it was time to switch to the specific disciplines. So here I asked it to create new case studies for each of the undergraduate disciplines that the university has.

These are much better.  Now, keeping these approaches in mind and the policy, create a new case study for each of the following disciplines:

[Enter all relevant disciplines of the school]

Prompt #4: Deepening the case studies

Now, it produced them but as often happens when working with large numbers, ChatGPT will keep it at the surface level. This prompt had it revise and go into more depth as well as add questions for the faculty.

Most importantly, I asked it to only do 4 disciplines at a time so as to get as much detail for each as possible. Were I to have more time, I would have does three distinct case studies per discipline. That would have significantly made it more interesting for each faculty member.

These are great.  Revise them one more time.  Expand the context of the case study to 3 or more sentences.  Change the last sentences that start with "the gray area" or "the challenge" and reframe these as 2-4 questions that should be asked to faculty who might be discussing these scenarios to figure out appropriate ways of navigating them with students.

Only revise these case studies:

[List 5 cases at a time]

You can the full results under the Case Studies section of the activity document but I’ll include two examples here.

Example Case Study #1: Child Studies

Students in a child studies program use AI to analyze vast amounts of data on child development and learning patterns. Based on this analysis, they propose new educational strategies and interventions. While the AI provides insights into effective learning models, the students’ role in creatively applying these insights to real-world educational settings is crucial.

1. How can we evaluate the originality and practicality of the educational strategies proposed by students when they are heavily informed by AI data analysis?

2. What criteria should be used to assess students' ability to apply AI-generated insights to diverse educational contexts?

3. In what ways can we ensure that students maintain a critical perspective on the data provided by AI, especially in the context of child development?

Example Case Study #2: Management Plans

In a management course, students use AI tools to simulate various business scenarios, including market expansion, crisis management, and employee engagement strategies. The AI provides a range of potential outcomes based on different management approaches. Students are then tasked with developing a comprehensive management plan based on these simulations. The focus is on how students synthesize AI-generated scenarios with real-world management theories and practices.

1. How can faculty assess the extent of students’ understanding and application of management theories in light of AI-generated scenarios?

2. What strategies can be employed to ensure students are not overly reliant on AI for strategic decision-making?

3. How can educators encourage students to critically evaluate AI suggestions and integrate ethical considerations into their management plans?

You can see there is some repetition and were I had more time, I would look to spend a good amount of time on each discipline and building out more meat and specificity. Still, as a starting activity for faculty in thinking about and engaging with the policy and their classroom, it has some good elements to it.

Now, what’s great about this activity is that individual faculty can play with these prompts for their own figuring out. A faculty member can edit these prompts to relate to just a specific discipline and ask it to come up with different case studies within that discipline.

What’s more the faculty member can then engage with the generative AI to get feedback based upon how the faculty member would answer or respond. One more layer of coolness to this is that if they have the app on the phone, they can do this as a dialogue and not have to type but think aloud in conversation with the AI.

The Guidance for the Faculty

With all that set up in advance, during the session, here’s how I proceeded.

First, I laid some simple ground rules.

1. Try them out

2. They won’t be perfect

3. It’s ok to adjust/adapt the prompts to suit your needs/wants

4. You can experiment but be sure to share back what you did and what you found

5. The goal is to learn more about different ways to use this

I wanted them to give this an honest try and also know that people will do their own thing or take it in the direction they want. I was game for this, but wanted them to share back.

The Directions

From there, I walked them through what to do and we worked through each step.

Step 1: Individual Work (8 minutes)

1. Review the case study in your discipline.

2. Write down your answers to the questions listed and anything else you feel that you would do in approaching the scenario.

Step 2:  Work with Generative AI (5 minutes)

1. Go to one of the following generative AI tools.

   1. ChatGPT

   2. Claude

   3. Perplexity (do not need an account).

2. Copy the following prompt (but where you see [ ], enter in specific information requested).

   1. You are an expert in teaching, assessing, and supporting students in [Discipline of the case study].  You are navigating the balance between what generative AI can do to benefit teaching and learning and also have concerns about how it might impede or obstruct the learning process.  I will provide you with a case study and you will provide 5 recommendations for how you would work with those students in that context.  Your recommendations should be in a table form that includes 4-5 sentences about each recommendation, a rating on 1-5 (5 the hardest) on how easy-hard implementation would be, and 3 pieces of advice on how you would revise or adjust your course the next time you are teaching it to prevent this.  Say, “Let’s begin” when you are ready for the scenario.

3. Review the results and choose one of prompts below to enter in:

   1. “Expand option [#].  In particular, provide a detailed explanation of how I would do that.  Include the necessary and helpful steps, the amount of time each step would take, why this step is important, what other resources that are out there that might help me with this.”

   2. “Provide 5 completely new recommendations in the same format.”

   3. “I have limited time and limited resources. Choose the easiest to instill option that also will effectively resolve the tension. Explain and justify your choice.”

4. Review the results and follow up with more questions if needed.

5. Enter the following prompt & review the results

   1. “Returning to the case study itself and the concerns, challenges, and opportunities with generative AI.  Please provide 5 different and distinct answers to this question. Ground your answers in different effective and inclusive pedagogies:  [Copy/paste the 1st question]”

6. Enter the following prompt & review the results

   1. “Returning to the case study itself and the concerns, challenges, and opportunities with generative AI.  Please provide 5 different and distinct answers to this question. Ground your answers in different effective and inclusive pedagogies:  [Copy/paste the 2nd question]”

7. Enter the following prompt & review the results

   1. “Returning to the case study itself and the concerns, challenges, and opportunities with generative AI.  Please provide 5 different and distinct answers to this question. Ground your answers in different effective and inclusive pedagogies:  [Copy/paste the 3rd question]”

Step 3:  Breakout Room AI (8 minutes)

1. In groups of 2-3, share the following

   1. What did you have that the generative AI did not?

   2. What did the generative AI suggest that you hadn’t thought of?

   3. What was most useful and/or most surprising in this exercise?

2. Be prepared to share back your thoughts (vocally or in the chat).

Review of the Activity

It went reasonably well based upon the conversation afterward. But as first-attempts go, it did not go perfectly. The first is as I mentioned the edge cases needed more details, variation, and choices for each discipline. I plan to build these out more robustly going forward. Ideally, I want more details and richness of the case study and make sure it feels distinct and different. I also want to offer more than 1 per discipline so that faculty have some choice.

The timing was off. This was a challenge. I wanted them to work on it a bit on their own and play with entering the prompts (they were notified they would be using the tool during the event). But I think more time was needed before getting them together. And, of course, group time is never quite long enough. This was also true for reporting back. I should have structured it differently to make room for more reporting back and being able to ask follow-up questions.

I’ll be doing something similar in another week and feel like I can build upon this. I’ll definitely share back.

AI+Edu=Simplified by Lance Eaton is licensed under Attribution-ShareAlike 4.0 International