Today, I am excited to present the first article of a new GPTZero blog series called "AI Conversations with Teachers." In this series, we will explore the implications of AI-generated content in different educational settings. For our inaugural conversation, I am chatting with Professor Christa Laser about the use of AI in legal education. Before we get into the conversation, for brief context, Professor Laser is an intellectual property law professor at Cleveland State University College of Law. She comes to teaching after nearly a decade of practice experience as an intellectual property litigator and has deep expertise in patents, trademarks, copyrights, false advertising, pharmaceutical litigation, and regulation, and technology law.

Could you please first introduce yourself and tell us all about who you are?

Hi, I'm Professor Laser, and I teach innovation law and intellectual property law at Cleveland State University. I made the switch to teaching after practicing for about 10 years, and I really love it. Currently, in my course on innovation law, I always have at least one session dedicated to discussing the developments in artificial intelligence. This past year, ChatGPT was definitely at the forefront.

How have you incorporated the use of AI generated writing tools like ChatGPT in your curriculum, so far?

I allow my students to use large language model-generated content as part of their writing process for their papers. Balancing the desire for students to learn how to write well while utilizing the new available tools can be challenging. But I do my best to convey to students that they are responsible for the output, and they can decide how to use AI assistance. I just ask them to disclose to me what they're doing.

Because you actively allow AI chatbot use, like ChatGPT, in your student's writing, have you noticed any change in writing quality as an effect of your decision?

As a teacher, I want the output to look like legal writing. But to sound like legal writing, it should sound scholarly and have a certain tone to it. That tone is very different from the tone generated by ChatGPT which tends to be a little bit more fluffy with a language that is not suitable for concise legal writing.

This year, [with my approval] only one student decided to use ChatGPT extensively. Helping students integrate AI-generated content into their writing can be challenging because they might not always realize the difference in quality compared to human writing. When reading it, you can feel that the content is distinct from other types of student-produced content.

When evaluating the AI-generated work, there was some back and forth with the student who heavily relied on ChatGPT, discussing whether it was fair for me to judge their writing differently when I instructed to them that they were allowed to use it.

I find it fascinating that only one student took up your offer to use LLMs to assist their writing process. Do you believe that students are intrinsically motivated to not rely on AI chatbots?

I think that there were students that were saying, "I don't want to start from scratch with something I'm not familiar with, I just feel a little bit more comfortable doing what I've always done with my writing." I think some students thought it seemed helpful, but didn't know the best way to edit the content down once you've generated content. With all the hallucinations, you have to be very careful with your prompt writing to get anything useful.

There are students willing to push boundaries and explore new technologies when permitted, and I encourage them to do so. However, most students are not in that situation. The student who used ChatGPT spent significant time doing prompt editing and checking the content for accuracy, which often takes more time than writing from scratch.

What has the general consensus been surrounding the use of ChatGPT in your work environment? Are there policies CSU is working on to preemptively mediate the use of AI for the coming school year?

I think that professors in the law school environment are very concerned about the use of ChatGPT to pass the bar exam. And this concern will only grow as AI becomes increasingly intelligent.

Initially, when we discussed AI's influence on our classrooms, ChatGPT couldn't pass the bar exam. Most of us structure exams similarly to the bar exam, so students can prepare to become attorneys. Now that ChatGPT can pass the bar exam, professors are worried that students may set themselves up for failure if they rely on such tools during the exam, as they won't be permitted to use them in a closed Internet environment like the bar exam.

What measures are you taking to combat the issue with ChatGPT on exams?

This semester, in the fall, I think I'm probably not going to allow the use of the Internet. I'm still going to make my exams open book and open notes, but I don't think I'm going to allow Internet usage because I don't want students to just take my exam, copy it into ChatGPT, and rely on those answers.

Are there any last thoughts on how you will be addressing AI in your classroom for the coming year?

I think having a really accurate detector would be wonderful. One thing I am particularly concerned about is that in law, there are so many AI tools that are being used without people knowing that it's an AI tool. So for example, Westlaw and LexisNexis, which are the tools that allow people to search for case law, are some of the tools that have tons of AI tooling built into their products, and many students think that it's written by humans.

I am hoping that students will be forthright when they use AI generated content. If you use ChatGPT, tell me your prompts and tell me what portions were written using AI. I think that information would give me everything that I need to fairly judge their work.

In the end, I want the students to learn how to write on their own. But I also want them to learn how to write with all the tools that are available.

Disclaimer: Interviewee responses have been lightly edited for clarity.