Lunch and Learn: From Plato to Pixar – Using Storytelling Frameworks to Drive Learner Engagement and Improved Outcomes

On Tuesday, October 8, 2024, the Center for Teaching Excellence and Innovation (CTEI) hosted the first Lunch and Learn of the Fall 2024 semester. Brian Klaas, Assistant Director for Technology and instructor at the Bloomberg School of Public Health, presented: From Plato to Pixar – Using Storytelling Frameworks to Drive Learner Engagement and Improved Outcomes. Caroline Egan, Teaching Academy Program Manager, moderated the presentation.

Brian Klaas opened the presentation by highlighting the valuable role storytelling plays in the classroom. While sharing facts is essential, building a narrative around those facts provides the context that gives them meaning. When we teach, we enact a narrative; we are telling a story. Stories help students connect abstract or complex ideas to real-world applications, making the content more relatable and engaging. Students are more likely to recall and retain information that resonates with them on some level. Storytelling also encourages students to think critically as they analyze characters, events, and outcomes, which can deepen their understanding of the material.

Klaas went on to describe the four cornerstones of narratives and encouraged audience members to think about how they might apply them to their teaching:

  1. Conflict: What problem are we trying to solve? What are we trying to overcome? What stories as a presenter can I frame through conflict? Conflict gives us drama and tension.
  2. Character: How can you draw people into your lectures or stories? We need facts and data, but we also need to engage with the emotional part of the brain. There are many opportunities to include characters/people into your data – including yourself!
  3. Segmenting: Organizing and chunking information, such as chapters in books, acts in plays, etc., gives our presentation a logical flow. We present one idea at a time, declare why it’s important, before moving on to the next one. Segmenting gives us a sense of moving forward.
  4. Reflection: What does this mean to me? Why is this important? As educators we can incorporate reflection into our teaching in various ways, including think-pair-share, self-assessment, journaling, etc.

Klaas acknowledged that it can be difficult to make the shift from sharing facts to telling stories. He introduced the audience to two storytelling frameworks that can help with this process: the Hero’s Journey and the Pixar Framework.

Circular diagram showing the 12 steps of the Hero's Journey.The Hero’s Journey is a popular template that follows the adventures of a hero who faces some sort of conflict, overcomes various obstacles along the way, and eventually emerges victorious. In the process, the hero is transformed in some way, embracing newfound knowledge and lessons learned. Examples of stories that follow this template include: The Lord of the Rings, Harry Potter, and Star Wars. Originally 12 steps, The Hero’s Journey template can be divided into 3 main acts:

  1. Departure/Separation: Where do things stand? This act presents the background, the current cultural context, followed by an event that pushes the hero to begin the journey.
  2. Initiation: What went wrong? What is the issue? Why can’t we move forward? This act explores the series of challenges faced by the hero that help to create tension and conflict.
  3. Return: What lessons did we learn? Why does this matter? This act shows what was learned and how the world is changed, why it is meaningful.

Klaas gave an example of a Hero’s Journey story:
He told the audience about one of his former students who traveled to Nigeria to study that country’s waste disposal and sewage system. While there, she noticed a woman repeatedly leaving her hut to go out into the grass. When the student asked why, she was told that the woman recently had a baby, was having some problems, and wasn’t able to engage with the community. The student learned that the woman experienced an obstetric fistula, a serious complication of childbirth that causes women to suffer incontinence, shame, and social segregation. The student went to community leaders and tried to explain to them that this is common, that the woman just needed proper care. The leaders rejected her plea. For the student, this was a lesson in cultural humility. She learned that local hospitals are not equipped to deal with this issue. This encounter led to her to shift her studies to help women deal with this problem. She ended up working on a project to help women in this region.

Klaas noted that the hero does not have to be a person – it could be a cell, for example – and that this framework could be a powerful way to present research to the world.

The Pixar framework, which forms the foundation of animated Pixar films, such as Finding Nemo and Toy Story, uses the following format:

  • Once upon a time… (context of the world)
  • And every day… (everyday life in that world)
  • Until one day… (incident that launches the story)
  • And because of this… (the character’s journey)
  • And because of that… (new journey the character takes)
  • Until finally… (resolution of the story)

This framework can be applied to a number of different areas, such as literature, health sciences, and history. It can be used to tell a story about why things fail or why things change – it doesn’t always have to be a success.  Klaas gave the example of how this format could be applied to narrate a problem such as antibiotic resistance:

  • Once upon a time… You get an antibiotic every time you go to the
    doctor
  • And every day… Antibiotic-resistant bacteria are on the rise,
    and here’s the data to prove it!
  • Until one day… Strategy for reducing antibiotic
    overprescription and sparking new research
  • And because of this…Reduction in antibiotic-resistant microbes;
    increase in funding for new antibiotic research
  • And because of that… Better health outcomes for people; better
    incentives for private research
  • Until finally… Generations of new antibiotics without living
    in fear of paper cuts killing us

Audience members were asked toOrange clownfish from Finding Nemo movie. share their ideas of how they might use this framework in their courses: One guest suggested using the framework to share research findings about high cholesterol with students in a concise manner. Another guest suggested using it to describe the history of drunk driving and the evolution of the breathalyzer test, leading to the development of a handheld alcohol detection device. And another guest suggested using it to talk about climate change in order to facilitate discussion.

The presentation wrapped up with a brief Q and A session:

Q: How vital is it to center our teaching around one protagonist as opposed to many? BK: Use storytelling where you can. It’s hard to use all the time, but having context makes it relevant. Always look for meaning, something that will be remembered down the line.
Guest: It helps when trying to make data meaningful. Seeing the story of one person works well. Sometimes too many people [as protagonists] is overwhelming, it becomes abstract. But sometimes showing the numbers affected [in terms of data] can be very powerful.

Q: Students have to write personal narratives for their medical school applications; they try to present themselves as the perfect candidate. Do you recommend using storytelling here?
BK: Yes, you can use storytelling in an application. Storytelling comes from written form, so yes, it will work in writing. My son is in the process of writing his college essays – he’s using the Pixar model right now. It made it easier on him to tell the story once he had a backbone.

Q: Could storytelling be used to help students solve math-based problems? Do you see this?
BK: I haven’t used it this way, but a colleague used it to share complicated math and statistics results from his research. I can see it working, through logical proofs. You don’t have to use the whole Pixar framework – you can use 4 of the 6 steps, for example, to make it work for you.

Amy Brusini, Senior Instructional Designer
Center for Teaching Excellence and Innovation
 

Image source: Lunch and Learn logo, Writer’s Digest, Disney/Pixar

Lunch and Learn: Generative AI Uses in the Classroom

On Tuesday, April 23rd, the Center for Teaching Excellence and Innovation (CTEI) hosted a Lunch and Learn on Generative AI Uses in the Classroom. Faculty panelists included Louis Hyman, Dorothy Ross Professor of Political Economy in History and Professor at the SNF Agora Institute, Jeffrey Gray, Professor of Chemical and Biomolecular Engineering in the Whiting School, and Brian Klaas, Assistant Director for Technology and instructor at the Bloomberg School of Public Health. Caroline Egan, Teaching Academy Program Manager, moderated the discussion.  

Louis Hyman began the presentation by reminding the audience what large language models (LLMs) like ChatGPT can and cannot do. For example, ChatGPT does not “know” anything and is incapable of reasoning. It generates text that it predicts will best answer the prompt it was given, based on how it was trained. In addition to his course work, Hyman mentioned several tasks he uses ChatGPT to assist with, including text summarization, writing complicated Excel formulas, writing and editing drafts, making PowerPoint tables, and turning image files in the right direction.

In Hyman’s course, AI and Data Methods in History, students are introduced to a variety of tools (e.g., Google Sheets, ChatGPT, Python) that help them analyze and think critically about historical data. Hyman described how students used primers from LinkedIn Learning as well as Generative AI prompts to increase their technical skills which enabled them to take a deeper dive into data analysis. For example, while it would have been too complicated for most students to write code on their own, they learned how to prompt ChatGPT to write code for them.  By the end of the semester, students used application programming interface (API) calls to send data to Google, used OpenAI to clean up historical documents and images presented using optical character recognition (OCR), and used ChatGPT and Python to plot and map historical data.Two maps of 1850 New England showing the number of congregational churches and the value of congregational property. Data points plotted by students using AI.

Hyman noted that one of the most challenging parts of the course was convincing students that it was OK to use ChatGPT, that they were not cheating.  Another challenge was that many students lacked basic computer literacy skills, therefore, getting everyone up to speed took some time. There was also not one shared computer structure/platform. The successes of the course include students’ ability to use libraries and APIs to make arguments in their data analysis, apply statistical analysis of the data, and ask historical questions about the results they were seeing in the data.

Jeff Gray continued by describing his Computational Protein Structure Prediction and Design course that he has taught for over 18 years. In this course, students use molecular visualization and prediction tools like PyRosetta, an interactive Python-based interface that allows them to design custom molecular modeling algorithms. Recently, Gray has introduced open-sourced AI tools into the curriculum (AlphaFold and RoseTTAFold), which predict 3D models of protein structures.

Example of protein folding using AlphaFold.

One of the challenges Gray mentioned was the diversity of student academic backgrounds. There were students from engineering, biology, bioinformatics, computer science, and applied math, among others. To accommodate this challenge, Gray used specifications grading, a grading method in which students are graded pass/fail on individual assessments that align directly with learning goals. In Gray’s class, students were presented with a bundle of problem sets categorized at various difficulty levels. Students selected which ones they wanted to complete and had the option of resubmitting them a second time for full credit. Gray is undecided about using this method going forward, noting that half of the students ended up dropping the course when they tried to complete all of the problems instead of just a few, and found the workload too heavy.  Another challenge was how to balance the fundamental depth of the subject matter versus application.  To address this, Gray structured the twice weekly class with a lecture on one day and a hands-on workshop the other day, which seemed to work well.

Brian Klaas teaches a one credit pass/fail course called Using Generative AI to Improve Public Health. The goal of this course is to allow students to explore AI tools, gain a basic understanding of how they work, and then apply them to their academic work and research. In addition to using the tools, students discussed the possible harms in Generative AI, such as confabulations, biases, etc., the impact of these tools in Public Health research, and future concerns such as the impact on the environment and copyright law. Klaas shared his syllabus statement regarding the usage of AI tools in class, something he strongly recommends all faculty share with their students 

Hands-on assignments included various ways of using Generative AI. In one assignment, students were asked to write a summary of a journal article and then have GenAI write a summary of the same article geared towards different audiences (academics vs. high school students). Students were then asked to analyze the differences between the summaries.Sample instagram post created using AI showing people from different cultures dressed as medical professionals. For another assignment, students were asked to pick from a set of topics and use Generative AI to teach them about the selected topic, noting any confabulations or biases present. They then asked GenAI to create a five-question quiz on the topic and take the quiz. A final assignment was to create an Instagram post on the same topic including a single image and a few sentences explaining the topic to a lay audience. All assignments included a reflection piece which often required peer review.

Lessons learned: Students loved the interdisciplinary approach to the course, confabulations reinforce core data research skills, and learning from each other is key.

The discussion continued with questions from the audience: 

Q: What would you recommend to an instructor who is considering implementing GenAI in the classroom? How do they start thinking about GenAI?
JG: Jupyter notebooks are pretty easy to use. I think students should just give it a try.
LH: I recommend showing students what ”bad” examples look like. The truth is, we can still write better than computers. Use AI to draft papers and then use it as an editing tool – it’s very good as an editing tool. Students can learn a lot from that.
BK : I recommend having students experiment and see where the strengths lie, get an overall awareness of it. Reflect on that process, see what went well, not so well. Feed in an assignment and see what happens. Use a rubric to evaluate the assignment. Put a transcript in and ask it to create a quiz on that information. It can save you some time.

Q for Brian Klaas: What version of GPT were you using?
BK: Any of them – I didn’t prescribe specific tools or versions. We have students all over the world, so they used whatever they had. ChatGPT, Claude, MidJourney, etc. I let the students decide and allowed them to compare differences.

Q for Jeff Gray: Regrading the number of students who dropped, is the aim of the course to have as many students as possible, or a group who is wholly into it?
JG: I don’t know, I’m struggling with this. I want to invite all students but also need to be able to dig into the math and material. It feels like we just scratched the surface. Maybe offering an intersession course to learn the tools before they take this class would be helpful. There is no standard curriculum yet for AI. Where to begin…we’re all over the map as far as what should be included in the curriculum.
LH: I guess it depends on what your goals are. Students are good at “plug and chug,” but bad at asking questions like, “what does this mean?”
BK: We didn’t get to cover everything, either – there is not enough time in a one credit class. There are just so many things to cover.

Q: What advice do you have for faculty who are not computer scientists? Where should we start learning? What should we teach students?
LH: You can ask it to teach you Python, or how to do an API call. It’s amazing at this. I don’t know coding as well as others, but it helps. Just start asking it [GenAI]. Trust it for teaching something like getting Pytorch running on your PC. Encourage students to be curious and just start prompting it.
BK: If you’re not interested in Jupyter notebooks, or some of the more complicated functions, you can use these tools without dealing in data science. It can do other things. It’s about figuring out how to use it to save time, for ideation, for brainstorming.
JG: I have to push back – what if I want to know about what’s going on in Palestine and Israel? I don’t know what I don’t know. How do I know what it’s telling me is correct?
LH: I don’t use it for history – but where is the line of what it’s good and not good at?
BK: I would use it for task lists, areas to explore further, but remember that it has no concept of truth. If you are someone who knows something about the topic, it does get you over the hurdles.
JG: You have to be an expert in the area to rely on it.
LH: Students at the end of my course made so much progress in coding. It depends on what you ask it to do – protein folding is very different than history that already happened.

Q: How can we address concerns with fairness and bias with these tools in teaching?
BK: Give students foundational knowledge about how the tools work. Understand that these are prediction machines that make stuff up. There have been studies done that show how biased they are, with simple prompts. Tell students to experiment – they will learn from this. I suggest working this in as a discussion or some practice for themselves.

Q: Students have learned to ask questions better – would you rather be living now with these tools, or without them?
JG: Students are brainstorming better. They are using more data and more statistics.
BK: AI requires exploration and play to get good responses. It really takes time to learn how to prompt well. You have to keep trying. Culturally, our students are optimized for finding the “right answer;” AI programs us to think that there are multiple answers. There is no one right answer for how to get there.
LH: Using AI is just a different process to get there. It’s different than what we had to do in college. It was hard to use computers because many of us had to play with them to get things to work. Now it all works beautifully with smart phones. Students today aren’t comfortable experimenting. How do we move from memorization to asking questions? It’s very important to me that students have this experience. It’s uncomfortable to be free and questioning, and then go back to the data. How do we reconcile this?

JG: What age is appropriate to introduce AI to kids?
LH: Students don’t read and write as much as they used to. I’m not sure about the balance.
Guest: I work with middle and high school teachers. Middle school is a great time to introduce AI. Middle school kids are already good at taking information in and figuring out what it means. Teachers need time to learn the tools before introducing it to students, including how the tools can be biased, etc.

Q: How can we encourage creative uses of AI?
BK: Ethan Mollick is a good person to follow regarding creative uses of AI in education and what frameworks are out there. To encourage creativity, the more we expose AI to students, the better. They need to play and experiment. We need to teach them to push through and figure things out.
LH: AI enables all of us to do things now that weren’t possible. We need to remember it’s an augment to what we do, not a substitute for our work.

Resources:
Hyman slides
Gray slides
Klaas slides

Amy Brusini, Senior Instructional Designer
Center for Teaching Excellence and Innovation
 

Image source: Lunch and Learn logo, Hyman, Gray, and Klaas presentation slides, Unsplash