A panini recipe – yes, the grilled sandwich – provided an unlikely source of inspiration for one UM-Flint staffer to delve into the world of generative artificial intelligence.
"I was with my dad in Kroger shopping for a Superbowl party when he suggested we make paninis," said Nick Gaspar, director of the Office of Online & Digital Education. "I remember hearing about this AI tool that talks back to you. So I asked ChatGPT and it came back with a panini recipe and shopping list. After the weekend, I came back to the office and said, 'This is it, this is what we're doing now."
While Carson Waites, ODE senior instructional designer, had made Gaspar aware of the tool months before, it took that real-world application to put the power of generative AI in clear relief for him. After months of research and tri-campus collaboration with Ann Arbor and Dearborn, ODE launched the Generative AI Prompt Literacy course. The course is free, delivered online asynchronously, and shares the hows and whys of constructing effective AI prompts to leverage this transformative technology fully.
What is generative AI?
As Gaspar explains in this video, AI broadly refers to a technology that imitates human intelligence to carry out various activities such as learning, problem-solving, and decision making. We've seen (or perhaps just experienced in the background) "narrow AI" at work for a number of years now. Narrow AI is programmed to accomplish a specific task – like recommending videos on YouTube or playing chess.
Generative AI is much broader in scope and defined by its ability to create new content. Text, music and images can all be created using generative AI. Some common generative AI tools include:
- ChatGPT. The most widely known tool, ChatGPT provides text responses to a wide variety of user prompts.
- Dall-E2. An AI system that can create realistic images and art from user input.
- AIVA. This tool uses generative AI technology to create original music.
- GitHub Copilot. Supports computer programmers.
- Scribe. A dedicated AI writing assistant.
Generative AI is nearly ubiquitous in its usefulness, but its true power can only be harnessed through effective prompting.
"For people not skilled at generative AI prompting, these tools can feel like a toy or a fad. For those who have been using them since the beginning, the results can feel like magic," Gaspar said. "We saw a divide forming between those who can prompt and those who can't."
Gaspar and his team saw that as the primary issue to address, and thus the course was developed.
Generative AI in education
New innovations create new opportunities, but also breed new concerns. The power of generative AI has the potential to transform work. As Gaspar explained, "you can automate out a lot of busy work and focus on what is truly important." But how should such technology be viewed in the context of K-12 and higher education?
The UM-Flint AI Task Force has strongly recommended against a blanket ban of generative AI in coursework. According to a task force document, "Banning GAI from coursework is not realistic. A ban is unenforceable, and it is impossible to detect all uses of GAI."
A more productive path forward is to first clearly define expectations regarding generative AI in a course syllabus. These expectations may include defining what tools are appropriate to use, in what manner it is acceptable to use these tools, and how students should give credit to AI should they use it.
Gaspar said that the reasoning behind these expectations is also important to share with students.
"Whether you are in the sixth grade or grad school, the common question is, 'why do I need to learn this?' We need to do a better job of explaining that 'why' and applying it and connecting it to their academic career," he said.
After defining expectations, the next step to productively incorporate AI into coursework is the creation of "authentic assessments." For example, if a health care student utilizes ChatGPT to create a diagnosis and treatment plan for a simulated patient case, part of that assignment might require the student to identify what the tool got wrong or didn't include in their response.
There are also a number of equity and accessibility issues that arise with the use of generative AI in education. David Luke, chief diversity officer at UM-Flint, said that generative AI "retains all of the biases of the information it intakes, including any stereotypes and misinformation present in the original text," so users must think critically when considering AI-generated information.
"Access to the technology needed for generative AI, as well as basic AI literacy, require a level of financial investment that some people cannot afford," Luke said. "The digital divide can contribute to increased socioeconomic inequity. That reality is one reason UM-Flint created its course and opened it widely to the public.
Enroll for free to increase your generative AI literacy
Since its launch in summer 2023, UM-Flint's Generative AI Prompt Literacy course has attracted more than 380 learners, including users from all three U-M campuses as well as Microsoft, Cornell University, Columbia University and more. The course is offered online asynchronously, meaning that it fits any schedule, and Gaspar said it takes an average of three hours to complete.
You can learn more about the course and sign up now online.
Logan McGrady is the marketing & digital communication manager for the Office of Marketing and Communication.