Experimental AI course will return this fall

English associate professor Abram Anders said the same thing to begin his experimental course, "Artificial Intelligence (AI) and Writing," every time it met this fall: "I'm so excited for today." The instructor was just as enthused as the 31 students to see what new things they would learn.

Ethics and AI

Anders and Dux Speltz developed a prompt at abramanders.com that can be put into any generative AI tool to act as an ethics tutor. Anyone can then ask the tool questions about AI ethics and it will present users with scenarios that may occur as they use AI. An AI in education checklist for students and instructors also is available.

With the semester complete and results reviewed, Anders is convinced this is just the beginning of what AI and curious students can achieve. Students appreciated the active learning and the tangible results that came from their expanding knowledge and success using AI tools, Anders said. All levels of undergraduate, masters and doctoral students were in the course.

Anders will offer the class next fall since he has received plenty of interest from students and inquiries about an advanced course for those who have already taken it.

"We will really look to grow to meet demand, whether that means adding sections or enrollment capacity," he said. "I was very excited to see them use AI as a tool to aid their education and not to replace their learning or do work for them."

Anders is not the only instructor in the English department looking to use AI in a class. Department chair Volker Hegelheimer is working with instructors in two core English courses (150 and 250) to introduce AI.  

The course

Anders designed the course to have a different focus each week, but most importantly students learned how to use different AI tools. Each AI tool required different prompting techniques.

"There is tremendous enthusiasm, but knowing how to use AI and get good results out of it, especially for situated applications it takes work and learning," he said. "Students quickly learned that you can't ask for something and take its first result. If you do, you will get mediocre outputs."

Beyond learning how the tools worked, students were asked to consider ethical issues inside and outside academic settings and how to correctly prompt each tool. The course primarily focused on large language tools, like ChatGPT and Bing AI, but also used image and video tools. Students pushed themselves and learned through weekly creative challenges using different techniques.

The course concluded with students doing a creative project, which Anders said he used as an indicator of the success and positive learning that took place. Postdoctoral research associate Emily Dux Speltz -- who helped Anders design, teach and review the course -- said the creative projects fit into five themes: interactive apps, creative multimedia, media production and design, writing assistance, and research and analysis.

"One student, a farmer with no prior coding or programming experience, developed a mobile app to aid sustainable farming practices," she said. "We were blown away by that."

Takeaways

Anders said he expects to see more AI courses across colleges as interest from both instructors and students grows. Dux Speltz said varying teaching styles -- such as lectures or letting students learn on their own -- prompted different but productive results.

"There was benefit to providing some core content around AI, but also providing the opportunity to have hands-on practice with those strategies right away was incredibly helpful," she said.

Anders also had students gather after using AI tools to share their thoughts and experiences. It was valuable to instructors and students because there wasn't enough time for everyone to sample all the AI options, Anders said. Each of the large language tools showed different strengths, leading students to use different ones for certain tasks.

Anders said success for instructors comes with plenty of trial and error before classes begin. He encourages instructors to spend time with AI tools they intend to use to ensure the tools  can produce what they will ask students to do. Instructor familiarity with the tools is just as important because AI will interact with content differently across colleges. Free versions of the tools can limit output, especially in media applications like videos. This may hinder some students if they are unable to afford pay versions of the tools.

"It is important to give your students some experience with the tool within the context of your course so they understand the things it can't do relative to the course material," he said. "Some kind of discussion about ethics and what kinds of things are and are not allowed is important."

Anders and Dux Speltz developed a prompt that can be put into any generative AI tool to act as an ethics tutor (see box).

The future

Anders said that as AI changes on a seemingly daily basis, faculty and staff need to be explorers as much as instructors, willing to pivot around AI for the foreseeable future. He believes it won't be long before AI becomes an incorporated part of many upper division courses.

"There will be dedicated AI courses in nearly every discipline," he said. "We have already seen AI and writing and AI and art and there will be an AI and … in almost everything for sure because it is already happening."