LLMs are swallowing the legal world. No, not the Master of Laws degree – I’m talking about large language models. These LLMs are here, they’re increasingly powerful, and they’re already being used by law students and legal professionals alike (with occasionally disastrous results). Even judges are getting on board. So, what does this all mean for legal education?
In ‘The ‘Other’ LLM: Large Language Models and the Future of Legal Education’, published in the European Journal of Legal Education, Vol. 5, No. 1, 2024, pp. 127–155, I reflect on whether LLMs could be used constructively in the law school classroom. I argue that LLMs could be like calculators for mathematics education – that is, despite fears, they could actually improve legal education if used correctly.
From Calculators to ChatGPT
When calculators were first introduced in math classes, there was widespread concern that they would make students lazy and hinder their ability to perform mental arithmetic. Fast forward to today, and calculators are an integral part of mathematics education, allowing students to focus on higher-level problem-solving skills.
I argue that LLMs could play a similar role in legal education. By taking care of some of the more routine tasks, these AI tools could free up students to engage more deeply with complex legal concepts and develop critical thinking skills.
The BAT Framework
In my article, I apply Simon Sneddon’s ‘Best Appropriate Technology’ (BAT) framework to assess the potential of LLMs in legal education. This framework provides a valuable lens for evaluating educational technologies. It consists of two key components:
- ‘Best’ refers to technologies that enhance learning, often by allowing students to focus on higher-order thinking skills.
- ‘Appropriate’ refers to technologies that can be implemented without detracting from the educational environment.
Looking at calculators via the BAT framework, it is clear that calculators enhance learning by bypassing time-consuming manual operations and providing instant feedback. They’re considered appropriate in classrooms as they don’t distract from the primary academic goal. Unlike more complex technologies, calculators are single-purpose tools that don’t tempt students with non-academic distractions.
The successful integration of calculators in mathematics education serves as a model for how LLMs might be productively incorporated into legal education. In the specific context of LLMs in legal education, I argue that these tools can be ‘best’ by bypassing time-consuming tasks and providing immediate feedback. However, their ‘appropriateness’ depends on how they’re used. When employed transparently to support learning, LLMs can be appropriate. But if used to replace students’ own effort entirely, or simply as a distraction, they become inappropriate. As such, the BAT framework underscores the importance of guiding students towards appropriate LLM use in legal education.
Aids, not Shortcuts
So, my analysis suggests that LLMs have the potential to be a BAT in legal education – but only if we guide students towards appropriate use. This leads to a crucial question: how can we encourage students to use LLMs as learning aids rather than shortcuts?
This question is at the heart of my article. My response: the best way to promote appropriate LLM use is to make legal education so engaging that students won’t want to outsource their learning to AI. Rather, they will want – or at least be inclined to – do the work themselves.
To that end, I propose that LLMs themselves can help us create the kinds of active, participatory learning environments that would encourage student engagement. In the article, I provide some examples of how LLMs could enhance different aspects of legal education:
- Discussion Preparation: Students can use LLMs to practice debating legal cases before class, helping them prepare for discussions and Socratic questioning.
- Discussion Prompts: LLMs can generate fresh, diverse prompts for class discussions, keeping both students and teachers engaged.
- Counterargument Generation: LLMs can present opposing viewpoints on controversial topics, encouraging broader class participation and combating ‘self-silencing.’
- Pre-reading Context: Students can use LLMs to gain background information on cases or legal concepts before diving into assigned readings.
- Post-reading Comprehension: LLMs can assist students in reviewing and discussing cases they’ve read, reinforcing understanding.
- Writing Prompts: Teachers can use LLMs to quickly generate diverse writing prompts related to course material or in response to in-class discussions, allowing for more frequent, in-class writing practice.
- Immediate Feedback: Students can receive quick feedback on short writing exercises from LLMs, promoting continuous improvement and ‘fast feedback.’
- Reflection Exercises: LLMs can guide students through reflective exercises about their learning, helping them connect course material to broader legal concepts and personal experiences.
There are many other possible uses, each of which can be used to make legal education more interactive, more personalized, and more engaging, all while promoting the appropriate use of LLM tools.
Changes are Coming
While I’m optimistic about the ability of LLMs to enhance legal education, the rapid advancement of these technologies means that changes to legal education are coming, whether we like it or not. And it is ultimately up to us as educators to shape how these technologies are used in our classrooms. Overall, I hope my article will spark discussions and debates among legal educators, students, and professionals. I’m particularly keen to hear from educators who have already started experimenting with LLMs in their teaching.
Keywords: Artificial intelligence, large language models, legal education
AUTHOR INFORMATION
Jack Wright Nelson is an Adjunct Research Fellow at the Centre for Banking & Finance Law, National University of Singapore.