On Wednesday, April 22nd, the Center for Teaching Excellence and Innovation (CTEI) hosted a Lunch and Learn featuring a student panel on Learning With AI: Student Voices on Studying, Research, and Writing. The panel featured Alisa Hamilton, a PhD student in Systems Engineering, Alisha Mason, a senior in Molecular and Cellular Biology, and Natalia Als, a freshman in Medicine, Science and Humanities. Caroline Egan, Teaching Academy Program Manager, moderated the discussion.

The panelists began by introducing themselves and describing their history with AI, including how they use it in an academic context. Alisa Hamilton uses it all the time for coding assistance and says it has definitely made her a better engineer. She does not use it much in her personal life but noted that she did use AI to help tile a bathroom. Alisha Mason was very “anti-AI” at first and did not use it much as a freshman because she was afraid of becoming too dependent on it. Her views have gradually shifted and she has since learned to use it as an effective study tool. Additionally, Mason uses AI as a “personal assistant” to help with tasks such as writing emails and revising her resume. Natalia Als was strictly “anti-AI” in high school, and continues to mostly avoid using it, but has noticed that it is more normalized at JHU than in her high school. She notes that instructors here include AI statements as part of their course policies, but she still does not feel like she has the skills to use it properly. She has observed peers using it in unethical ways and becoming too reliant on it. Unlike Mason, Als does not use AI in her personal life, preferring to take the time to write her own emails and complete other everyday tasks herself.

Caroline Egan asked panelists if they used AI in their learning, how has it been the most useful, and whether there have been instances in which it has hindered their learning. Hamilton finds it very helpful to upload PowerPoints or PDFs and ask AI to explain them, such as why certain parts of code work in an application. For her, this helps
“fill in the blanks” and she ends up getting more out of courses than if she did not use AI. Mason agreed, adding that ChatGPT helps explain concepts that she might not initially be able to grasp. She frequently has ChatGPT open as she’s listening to recorded lectures so she can ask questions. An ongoing hindrance is that the information from AI is not always correct, and she needs to check the sources. She says the pros outweigh the cons, however, and she plans to continue using it. Als described a situation where she was assigned a web-scraping exercise using AI and it produced a great deal of incorrect information. In the end, she had to go back and do the entire assignment manually. She supports those who use it and are still able to think critically, but says she has observed many of her classmates using it unethically usually because they are pressed for time. She says many freshmen lack time management skills and feel too overwhelmed to consider other solutions, therefore, they end up relying on AI to do their work. “They are not learning.”

Egan asked panelists if they thought AI was taking the place of office hours and if so, how could instructors entice students to continue attending. Hamilton suggested going through problem sets during office hours in a structured way. Mason says she does not attend office hours regularly, and that she always has ChatGPT pulled up when studying with friends. However, she does not find ChatGPT helpful with computational coursework, such as in Physics and Calculus, and feels AI cannot take the place of a tutor in these courses. In contrast, Als said she and her classmates frequently use Chat GPT in their Chemistry and Calculus courses. Even so, she finds the assistance too broad at times. As for office hours, Als says sometimes there is judgement from the TAs: “You don’t know this already??”  She suggested possible training for TAs so students would not feel intimidated during office hours.

Egan continued by asking what things faculty have done to help guide students in their use of AI, and what might they want faculty to know about AI from the student perspective.  Hamilton has only taken one class where AI was encouraged and feels there wasn’t enough guidance. She said it is a “slippery slope” asking students to use AI, but then to call them out for using it too much. In her experience, students do not always realize that they are cheating. She says faculty should be super-clear about what they expect, model what they want from students, and tell them explicitly what is not ok. Mason described a biology research course in which the instructor allowed students to use AI selectively on their projects and provided guidance on when its use was appropriate. Students grouped each step of their project into one of three categories: repetitive, editing, or facilitating. These categories helped determine whether AI could appropriately support a particular task. Mason noted that the exercise was especially valuable and continues to influence how she thinks about using AI in other courses. Als added a few suggestions: being open and honest with students, providing examples for them, and establishing a policy from the beginning and sticking to it. She appreciated one class where students were given the opportunity to create a policy collaboratively.

Egan raised the concern that instructors have with academic integrity violations when using AI. She asked panelists if this is a valid concern and how might instructors counteract those violations.  Mason acknowledged that this is a valid concern and suggested that instructors emphasize to students that they are here to learn how to think critically. She also suggested possibly using fear to dissuade students from cheating. Als agreed that her peers don’t seem to grasp that they are here to learn and “not just press Submit,” but is not sure that fear is the best solution. She suggested having conversations about academic integrity in smaller groups among peers, such as during TA sessions. Hamilton noted that this is a good opportunity for instructors to discuss what academic integrity means and provide examples for students who may be unfamiliar with the concept.

Several questions from the audience followed the discussion:

Q: I’d like the students to share their thoughts on faculty using AI to assist in teaching a class. Are you comfortable with instructors using it?  If faculty disclose their use, is that enough?

AH: I think disclosure is important. I would be bitter if I knew assignments were graded by AI solely. But I also know that TAs do most of the grading anyway.  I feel like I can have AI grade my work myself, but that is not why I’m taking this course.
AM: Multiple choice is ok but for free response questions, I prefer human eyes, human feedback. Instructors are able to see how students come to certain conclusions. I would want instructors to regrade if there was a question about the response.
NA: As long as the professor is prepared for regrade requests and questions, that’s fine. With free response, I would prefer to have my work graded by a human since that’s part of the grading process. What about other parts of the class? Will AI be creating the rubric? The course material? Where do we draw the line? You need context to grade. Are you giving AI enough context?
AH: I think faculty can use it the same way as students. It’s ok to draft an email or syllabus. And it’s a good idea to disclose that.
AM: I would feel weird if I knew AI generated the whole exam. I’m not sure why, but it takes away the creativity.

Q: As a faculty member, I take it personally when students use AI. I put so much work into commenting on writing assignments. It feels disrespectful to my time. I’m giving feedback at a high level which is the level they should be writing, but they are not understanding it. I’m wondering if it’s a double empathy scenario – all parties not understanding each other. Any ideas about this?
AH: If I thought a student was cheating, I would start with talking to that individual student.
AM: For writing courses, I would go old school to gauge students’ writing style and abilities. Use pen and paper – no technology – to see where students are and establish a baseline. Maybe a homework assignment where they write things down.
NA: We used a blue book in one of my classes to establish a baseline. Now the rest of the assignments are online, but the instructor knows how we write. It would be foolish for students to start using AI in that class now, as it would be so obvious. I think having one on one conversations with individual students is best.

Q: Is it invasive for professors to examine version history, such as using Google Docs?
NA: I understand the need and I personally do not feel like it is invasive.
AM: Students will find a way around version history if they want. I think starting with pen and paper is best.
AH: I recommend starting with an explicit agreement about what is expected, including what is and is not considered cheating.

Q: Do you use other AI tools used besides ChatGPT?
AH: I use Cursor and Claude Code, which are paid for by JHU.
AM: I use ChatGPT, Claude, and Gemini. I do not use HopGPT. I do not pay for any of them.
NA: I have used HopGPT and ChatGPT.

Q: Are you seeing a falling off of peers joining you in study groups because they are going off on their own to use AI? Or has it stayed steady while you use AI in the group?
NA: We are still meeting up and some people have ChatGPT open, but we’re still having social interaction. I am not sure about outside of my group.
AM: People who are going to study independently are going to study that way regardless of AI. Office hours for computational classes are always full. I still see study groups and my friends and I get together before exams. We always ask a TA or a professor if we have questions about an exam. I don’t really see it [falling off].
AH: We study outside of the lab to try out new AI tools.

Q: How do you and your peers think about the environmental impact, human rights issues, and privacy issues of using an LLM? How does this play into people’s decisions to use it/when to use it?
AM: We are all fully aware of the environmental concerns. Some students won’t touch it for this reason. Some are cautious about privacy concerns. My biggest fear is overuse. I need to know how to learn how to think.
NA: People are aware of the environmental impact. I see it posted on social media a lot. I’m from Chicago and there is a big data center about to go up there, which is concerning. People are not going to care enough until it actually affects them.
AH: I agree with all of these comments. We are going to end up with data centers on the moon, seriously. In terms of copyright, I make sure not to upload to anything that does not have a paywall.

Q: Do you think using AI reduces the depth of learning at all compared to before AI?
AH: No, I know what I want to learn and what I want to get out of it.
AM: It depends on how you use it. I feel like I use it to increase my depth of learning. For me, it is very efficient for troubleshooting. But we joke that there are so many people who overuse it and are going to have to rely on it for everything. It’s interesting to see underclassmen, how quickly they turn to AI for help. Upperclassmen take time to think, ask a TA. This is my observation.

Q: How do you feel about fairness, such as someone receiving the same grade as you by using AI?
NA: I think it goes back to an established policy that everyone is graded the same way. It has been frustrating when we know others have used AI, and we have not. I feel like I am being penalized for being honest and this person is not because they have submitted a perfectly written AI generated paper. I am not sure what else faculty can do. You don’t want to accuse people left and right.
AM: We will all have to go into professional fields and need to know certain things and how to communicate. Some people do not value this. Maybe faculty could emphasize this: “You are going to have to go out in the real world and you are not always going to be able to use AI.”

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

Image source: Lunch and Learn logo, Unsplash