by: Colin Black & Dyane O’Leary

1/8/2023. “ChatGPT is taking the world by storm. Let’s play around and figure out what all the fuss is about.”

7/8/2023. “We need policies for ChatGPT. Let’s share new syllabus language and brainstorm curricular changes.”

1/8/2024. “Good morning students. Welcome to our first ever course about Generative AI. Let’s get started.”

 

Time flies. About a year removed from the first ChatGPT boom of excitement, the treacherous waters have calmed–but the sea change is well underway. Legal educators wrestle with whether, when, why, and how to incorporate–or ban–Generative AI tools such as ChatGPT in classrooms and clinics and for assignments and exams. They’re commenting, warning, imagining, theorizing, worrying, and speculating.

We’re doing. Suffolk Law’s Legal Practice Skills program and Legal Innovation & Technology Institute steered head on into the storm, supporting one of the first law courses in the United States solely on Generative AI. Generative AI for Lawyers was a 2 credit, one-week intersession full-day in-person elective in January 2024 with 25 upper-level students. Designing and teaching a new course is one thing; staying afloat amid a new technology with daily updates, emerging ethics guidance, and changing diverse platforms is another. Students learned a lot. So did we.

For future navigation, we offer some reflections as others plot their own courses:

 

• Hands-on & High Level: Copyright Infringement. FTC inquiries. AI Bill of Rights. What about reading the terms of service and trying out prompting personas? Skills teachers to our core, we couldn’t teach this course without a heavy experiential tilt. Introduce a skill, practice it, discuss it, assess it. But emerging substantive legal issues with GenAI couldn’t be ignored. Our unique one-week format meant we couldn’t cover every base, but we aimed to at least touch on many. Several “sandbox” sessions created plenty of time for hands-on experimentation and discussion. Daily reflections and a final day dedicated to “law of technology” topics left us satisfied that students saw both the narrow trees and the bigger picture forest. Nowhere is the so-called skills/doctrine divide more irrelevant than on this emerging topic. Students deserve both.

 

• Tools of the Trade: ChatGPT is the Kleenex to the generic tissue. But Coke is not Pepsi, and ChatGPT is not the only tool in town. Far from it. Students worked with ChatGPT3.5, 4.0, Perplexity, Google Bard, Microsoft Bing, and Claude. Differences in platform features and outputs (e.g., some accept attachments or link to web sources while others don’t) were valuable food for thought. Could some exercises be streamlined if everyone uses the same tool? Sure. Will this list of tools probably be outdated days after we write this? You bet. Will every interaction with GenAI be consistent and exist in a vacuum? No. Are there material differences across platforms? Indeed. We embraced that diversity. Doing so sparked students’ appreciation of the importance of careful tool selection as soon-to-be modern consumers of legal technology.

 

• Prompt Away: The myth of the digital native in the context of GenAI is that students must be great promptors because they’ve grown up as Googlers. Not so. Massaging and iterating inputs for better, more nuanced quality outputs is new. It’s new to us; it’s new to them. To bridge this gap, our curriculum included prompt crafting workshops, sessions for analyzing and critiquing prompts, scenario-based assignments pitting different prompting techniques “against” each other, and opportunities for students to reflect on this new practice skill. We even had fun with “emotional prompting” instructing GenAI tools to follow the same advice we give ourselves:  take a deep breath and slow down!

 

• Blur the Legal v. Non-Legal Line: LexisAI, LawDroid, CoCounsel. But what about baking, stream freshwater fly fishing, and Taylor Swift song lyrics? There was great value in starting with non-legal tools and non-legal prompts. It gave students a foundational understanding of the predictive, statistics-based nature of how GenAI ‘creates’ the next word in simple and approachable personal contexts. By using tools like ChatGPT and DALL-E for non-legal queries, students got their feet wet in a low-stakes setting and realized that evaluating GenAI outputs is a task requiring knowledge and skill–even more so in the high-stakes context of lawyering. After all, if a user knows nothing about freshwater fly fishing, how do they realize ChatGPT’s tips actually involve trolling for saltwater bass instead?

 

• Strength in Numbers: Group work promoted a collaborative environment in a class filled with both novice beginners and tech-ies. Team-based learning facilitated peer-feedback and bonding and helped alleviate some “I’m not a tech-ie” imposter syndrome. Anyone who has played around with GenAI knows the possibilities are–truly–endless and diverse beyond our wildest imagination. With almost every prompt, exercise, or example, students were exposed to new angles and ideas. For more experienced students, group work provided a terrific opportunity to practice being the teacher themselves:  guiding beginner users as they’ll no doubt do upon graduation when they bring their new skills into the real world.

 

• Many Hands Make Better Work: Guest speakers enriched the course. Our expertise is in teaching–not computer science–and making complicated topics more accessible to students. Students know GenAI is a hot topic; they craved “hot off the press” perspective. We welcomed a computer science coder/lawyer colleague, our legal ethics expert dean, vendors building law-specific tools, and a technology business consultant. From neural network basics (not exactly a light dinner table topic . . . !) to trade secret business concerns, it was the perfect complement to deepen students’ exposure. The bonus? We learned too. And modeled for students the importance of upskilling throughout one’s career just as we did alongside them amid this new landscape.

 

• Diligent Digital Natives: Early academic GenAI chatter reeked of distrust. This course (albeit with 2L and 3Ls) suggests that fear is unfounded. Students critically engaged and were skeptics–balanced with a healthy dose of “wow, these tools are amazing.” With guidance, they weeded through hype and laid the groundwork for their own pro/con evaluations. Optimism about GenAI’s potential to enhance practice efficiencies never replaced recognition of the need for human judgment and oversight. Will some use GenAI to cheat? They might. Will some use it “wrong” in practice? They might. But that’s the user–not the tool. On the whole, we left inspired about how this small case study suggests most future lawyers will use GenAI–responsibly.

Anything new brings risks but offers rewards. In the spirit of starting somewhere, we found an imperfect yet perfect beginning to solving the “GenAI + legal education” equation. GenAI is moving fast. So too must legal educators if we’re going to keep charting the best course.