Education Technology Office hosts two workshops about classroom AI

Eliana Buono / THE GATEPOST

By Francisco Omar Fernandez Rodriguez Arts & Features Editor By Raena Hunter Doty Staff Writer The Educational Technology Office (ETO) hosted two workshops aiming to empower educators to use artificial intelligence (AI) in their teaching. The first workshop, held April 8, was called “Guiding Students to Create and Explore with AI.” It hosted Jennifer Dowling, professor of art, and Steven Courchesne, director of academic technology. Dowling presented on her RAMS101 course taught in the Fall 2024 Semester, which was called “Real or Digital? Creativity and Artificial Intelligence.” She said the class taught “creative exploration and practical approach to combining pixels and paint” using AI and traditional creative methods. This included fine art as well as other creative outlets, such as graphic design and creative writing. Dowling said she proposed the course after a sabbatical leave in spring 2022, which she spent exploring the ways AI and art can intersect with one another. Topics in the course included exploring how AI can improve creative workflow, using it for research, and idea generation. Dowling said some students were hesitant to use AI, either because they had been previously disallowed from it in high school or because they had ethical concerns with using the technology. “They were concerned. ‘Is my creativity going to be stolen?’” she said. “However, being the incurable optimist, even though I’ve been with technology for many years, I said, ‘We need to be encouraged,’ or at least I encourage them to try to put aside their tendency to be hesitant and just have that sort of healthy skepticism in terms of writing and learning and creating.” Dowling covered several different projects that she assigned her students, including different examples of the finished products. The first was a logo project where the students had to create a logo for a travel agency, either real or imagined, then attempt to prompt AI to create the same logo that they had. “What I wanted to convey to [the students] was that graphic designers are using AI. Yes, it could be considered a cheat or a workaround, but it also does help with those mundane tasks, so it enhances creativity,” she said. “It can help facilitate the process. It doesn’t replace it, and there’s a lot of concern that, ‘Oh, graphic designers aren’t going to be needed,’ but there are still valuable aspects to human comprehension and creative processes.” In comparing the student-made logos to those generated by AI tools, Dowling said the AI tools - especially those not made specifically to help design logos - could sometimes struggle to make viable designs, and often the designs wouldn’t resemble those made by the students. One student, who had designed a logo for Expedia, said the AI-generated logos were “less simplistic than the one I made and seemed more like movie posters than logos.” The next assignment she covered asked students to use a story posted on StoryCorps, a website hosting many short, non-fiction stories about American lives. In the assignment, students had to hand-draw a storyboard based on a StoryCorps story, then feed a summary of the story into an AI model that would expand on the story. They were asked to reflect on how well the AI model expanded on the story, how it compared to their original storyboard, and what they liked and disliked about the new story. While reflecting on the assignment, one student said, “I feel like if you wanted to come up with something original, AI would not be a helpful tool. However, if you just need to get a plot in mind and are not worrying about originality, AI would be helpful.” Other students reflected that using AI “simplified the process” and added more detail to the story to make it come to life. Finally, Dowling shared a few students’ final reflections over the course as a whole. One student reflected that, though they “do not really like generative AI and will probably still avoid using it,” the course helped them understand how it is not replacing human-made artwork and can facilitate creativity. Dowling acknowledged some of the ethical questions surrounding use of AI, and said these new technologies allow people to use and mimic the art of others in entirely new ways, which presents concern for many about copyright and intellectual property. “And one last note is that I remember when digital imaging was just starting to enter the art world. It wasn’t considered an art form, just like when photography was introduced, so there’s always going to be change or adaptation to new technology,” she said. After her, Courchesne presented his work in developing what he calls “thought experiments,” small lesson plans that allow teachers to integrate AI into their lessons and deepen student understanding. He said his thought experiments can be used to help students think more deeply about the past, future, or historical and fictional characters. Courchesne gave two examples of thought experiment lesson plans. The first was called the “AI Misinformation Campaign,” from the AI Pedagogy Project at Harvard, in which students had to prompt AI to generate a misinformation campaign. He said the lesson was geared toward helping students develop critical thinking skills around persuasive articles and misinformation. By understanding and analyzing the way AI models use rhetorical devices for information they know is wrong, students may be better suited to assess sources. Courchesne added it can also help better understand what AI can and cannot do - specifically, though many AI tools are programmed to correct information, they may still be used in malicious ways, and this assignment can demonstrate that. The second thought experiment was a roleplaying activity, where Courchesne showed how generative AI tools may be used to create personal experiences with historical or fictional figures. In his example, he roleplayed with the titular character from “Hamlet.” “I was curious to know - could this be a way that we could engage with a play, for example, or with other kinds of creative works?” he asked. Courchesne said he asked a few different types of questions - first factual questions that can be tested for right or wrong, then questions that were “adjacent” to the play, such as advice for what to do if he suspects his uncle stole from his father. He said the answers generated by the AI model seemed very in character for Hamlet, and overall considered the test successful. For anyone interested in doing this thought experiment on their own, Courchesne said it’s best to use a well-known character or historical figure, as more obscure options may not get the best results. To end, Courchesne directed educators toward Harvard’s AI Pedagogy Project, which he said contains many valuable resources for educators who want to integrate AI into their classes.

Adrien Gobin / THE GATEPOST

The second workshop, “(Easily) Create an AI Powered Classroom Chatbot For Your Students With Google NotebookLM,” was held April 9. Kevin Kennedy, instructional technologist, led the short seminar. NotebookLM is a Google product that can be used to create a chatbot, he said. It can be useful for students and for instructors. It does require signing in with a Google account, he added. NotebookLM can be used to make an online notebook for students “for when they have questions about class at 2 in the morning and you are not awake at 2 in the morning,” Kennedy said. In his example, he had a notebook already set up. He had uploaded a syllabus and two big assignments onto it, he said. The chatbot in the notebook only knows answers based on the uploaded documents, he said. He asked it when a specific assignment was due, and it gave him the due date and cited its sources. If it doesn’t know the answer it’ll say it doesn’t know, though it might add that the sources may indicate something, he added. While this already makes NotebookLM a great tool, “this is the least of what it can do,” Kennedy said. It can also be used as a course reference, he said. In another example, he had a notebook prepared for an astronomy course. Textbooks and chapters were uploaded, he said. It was able to answer multiple questions about astronomy asked by the audience. It was not able to answer a question about the best flavor of ice cream though, saying that its sources did not contain relevant information. The notebook can create a timeline, he added. It created a general timeline of astronomy, based on the provided information from the sources. It can also make a mind map, he said. It categorizes all of the sources into sub-contexts and shows it as a map. Clicking on a sub-context generates a question about it straight into the chatbot. It is useful for large sums of information, such as a whole subject, he added. The study guide feature creates a review of key concepts, Kennedy said. For smaller subjects, NotebookLM can be useful in different ways, he said. It can create a podcast, he said. The one he had as an example was 42-minutes long, but usually they’re 10 to 20 minutes. It can only create one downloadable podcast per day for everyone in the free version though, he added. Different roles can be assigned to determine what people can do. For bigger concepts though, the podcasts can focus more on the big picture and not cover the smaller details that students are more likely to need, he said. It can also create a frequently asked questions page, he added. Students can create their own notebooks to organize their thoughts and compile research, he said. “I’ve also found that sometimes it makes connections between sources that I myself would not make,” Kennedy said. NotebookLM can also be used in noneducational ways, he added. “I used this in ‘Dungeons & Dragons’ all the time,” Kennedy said. There is a limit on how many people can be added to the notebooks though, he said. Also, because this is a Google product it follows Google’s privacy policy, he added.