Artificial Intelligence (AI) is not new. The rise of ChatGPT, Google Bard, New Bing, and others in the academic space, however, is skyrocketing. My initial encounters with this rising AI were biased. As I scanned topics like academic integrity, academic dishonesty, and plagiarism, I quickly adopted others’ persuasive opinions based on limited information.
A few months ago, however, I attended a presentation that gave me a more complete view. Students were already using it as a tool, so I opened my mind to understanding what purpose AI served in the higher education classroom. Most tools are not innately good or bad; what we do with tools creates the perception and interpretation. I became curious. What were the pros, what were the cons, and how could I embrace the already organic shift disrupting higher education?
I created this article not to discuss potential legal and ethical issues surrounding AI, but rather to share my observations from a broad perspective.
What purpose does AI, such as ChatGPT, Google Bard, and New Bing, serve in the higher education classroom?
I found that both students and faculty have several common uses:
- Save time: With more non-traditional students entering the higher education space, they are pulled between multiple life demands. If there is a free tool that helps improve efficiency, why wouldn’t they use it? The time-saving tool’s abilities are not always student-centered though, as faculty can also employ AI to save time.
- Edit papers: Students can ask AI to edit papers. Beyond basic editing, students can add specifications such as editing for English as a first language, or editing for the academic environment. This utilization is often compared to utilizing Grammarly, or meeting with a writing tutor. Like Grammarly or writing tutors, students can take time to learn from AI’s corrections and suggestions. Instructors can use AI for editing as well. AI can edit many types of written documents.
- Generate outlines to facilitate linear communication: For non-linear thinkers, AI can save hours of becoming lost in thought. When non-linear thinkers can describe a concept and offer key words, they can prompt AI to help them create outlines. AI can further be directed to create outlines such as a PowerPoint deck. While the free versions do not generate an actual PowerPoint, they do create the outline to expand upon. The outline format helps improve linear communication and decrease confusion.
- Generate ideas: Writer’s block be gone! When writer’s block prevents content generation, AI can be asked to expand upon ideas. This expansion can help users develop ideas to include or research further.
- Generate complete content: AI can be directed to generate content such as a research paper on a specific topic. Users can specify parameters such as word count, number of resources, and types of resources. The longer the content, the more the content generation begins to break down; in dissertation length papers, thoughts often become contradictory and poorly organized. AI can also generate short content, such as answers to take-home quizzes or un-proctored multiple-choice exams. On the faculty side, AI can generate grading rubrics and well-stated course objectives in line with specified learning domains and Bloom’s taxonomy.
What are the pros of such uses?
Pros include saving time, enhancing readability, expanding vocabulary and grammar skills, improving linear communication skills, enhancing creative thinking, and uncovering perspectives from which to research or critically appraise topics further.
What are the challenges of such uses?
Some of the biggest challenges are preserving the actual learning experience and academic integrity. In reflecting on the learning experience, I found that the learning experience had morphed. Students and instructors were learning to ask new questions in new ways. The instant responses were shaping the nature of the communication and learning experiences that followed. I embraced this a novel learning process.
In the past if a student asked me about a topic like exercises for hip mobility, I would encourage them to use all of their resources. If they wanted to show me something they learned on TikTok, Instagram, YouTube, or somewhere else (even offline), I encouraged them to do so. Then, we would utilize our academic background to analyze the exercise. What are its pros? What are its cons? What structures are involved? Who is this exercise for? When is this exercise appropriate? When do risks outweigh benefits? The media they were interacting with every day added relevance and depth to our critical thinking skills, which enhanced both learning and contemporary work place applications.
I’ve embraced AI in the same way. If the students want to ask AI, all I ask is that they share what they find. From there we have an open critical thinking discussion on how our findings and current evidence present synergy, learning gaps, or discrepancy.
Academic integrity has presented a larger challenge, however. I found that some students were using AI for content generation. This was often for written assignments or take-home quizzes and tests.
A fellow instructor suggested I try audio or video presentation assignments instead. The result was the same, as I could tell students were simply reading from a pre-written source. This solution did not work in the classes I was teaching.
Identifying AI generated content continues to present a challenge. The AI generated content by-passes common plagiarism screening tools like Grammarly or SafeAssign. These tools compare percentages of the submission to pre-generated content already on the Internet. Since AI is generating novel compositions on an ever-changing algorithm, these tools are not picking up the novel content. The AI is also self-protecting, as it does not identify its own content if asked. If two different users enter the same prompts or content, it also changes its answers.
The four ways I have successfully identified AI generated content to date are:
- AI may generate resources that do not exist. When I examine the reference list, publication dates, authors, locations, or titles within each reference may not match up.
- The content lacks emotional elements. AI generated content tends to be very linear and logical. While I do have a few students with this natural writing style, when I read a paper that sounds like Spock from Star Trek wrote it, I start combing the references for inconsistencies.
- Students may display intra-personal writing inconsistency. If I know my class will have writing assignments, we start with a few in-class samples both as formative assessments for just-in-time teaching methods, and also as a student writing style baseline.
- For example, I may prompt students to start class with a one-minute brain dump on which topics to date they understand well, and which they would like me to review in more depth.
- As another example, if we have a paper coming up, we might do an in-class, five-minute brain storm. The prompt might be “Take five minutes to sketch the introduction for your paper.”
4. Students may display remarkable test and quiz score variation. When students score 100% on take home quizzes and tests, but score much lower on proctored exams, we have one-on-one meetings to brainstorm discrepancy sources. Another solution to take home test integrity employs technology such as the Respondus lock-down browser available at some institutions.
When AI content is suspected, I also have to check my assumptions. Instead of jumping to conclusions, I have a one-on-one meeting with the student. We try for an in-person meeting; if that is not possible, we try for a video meeting, as the student’s non-verbals provide just as many clues as the verbal expressions. I utilize intra-student submission discrepancies as a communication tool. We start with open-ended neutral questions such as “How is this class going for you?” We slowly narrow our conversation into their recent trends and possible discrepancies. From here, we hone in to brainstorm possible reasons for discrepancies. We also cover topics like academic integrity and what that does and does not look like in the academic space. Regardless of what the conversation uncovers and where it leads, it often concludes with “What can I do to help support your success?”
I share this article as an observation of how we are embracing AI’s existence as a tool in our higher education classroom currently. Uses and regulations will likely continue to develop. As a call to action, I would encourage those not familiar with AI to try it and see what type of responses it generates. Awareness is the first step in understanding a tool that is likely to exist in some capacity for decades to come.
Dr. Meredith Butulis comes from a background in education, business, physical therapy, and wellness/fitness coaching. She is currently an associate professor in the physical therapist assistant program at the State College of Florida. Prior to joining the State College of Florida faculty, she had been in physical therapy clinical practice, various wellness and fitness coaching roles, and an exercise science instructor and program dean. Dr. Meredith’s passions are sharing educational opportunities, and continuing her own journey in life-long learning.