Generative AI has the potential to achieve a lot of magic in legal education.
The educators can get a lot of assistance and new ideas, and the students can go further in their assignments – in the same limited timeframe.
And I say that as someone who went to law school and now teaches a small, data-driven legal design class called Modern Lawyers at Faculty of Law, Charles University in Prague.
Below is a personal reflection on how I use Generative AI in legal education.
Generative AI in my classroom
I am convinced that impact of Generative AI will be significant and overarching – in the legal industry and beyond.
So I make an active effort to explore the various ways it can add value – in both my preparation and in student assignments.
My key objectives are the following:
- Create space for safe and productive exploration
- Establish conditions to experience, not just study the technology and its limitations
- Facilitate constructive and informed debate about the suitability of the technology in various scenarios
How I approach teaching with Generative AI
I choose to treat Generative AI and prompt engineering not as separate topics but as an integral part of working with legal use cases and in performing assignments.
This is the new business-as-usual, so it needs to be approached as such. For example:
- Are we talking about legal design? We will use GenAI to work with user personas or to prepare service blueprints.
- Deep dive into legal ops? We will talk about the impact of GenAI-powered chatbots or augmented knowledge management
The students learn all the tricks by doing, not necessarily by instruction. Being nudged to work with Generative AI every week for three months creates the perfect conditions to develop creative prompting confidence.
How Generative AI influenced my classes
Generative AI can empower students to get further
Generative AI lowered the barrier for building cool stuff – in terms of tools and time constraints. Some students used Generative AI to create fully functioning tools. And countless others leveraged it for specific tasks.
You still need to have the ideas, make the tough design decisions, and have good information architecture and strategy. But you won’t get stuck because you don’t know how to start building.
I leave it completely up to the students to choose how they do the projects. I suggest some software (mostly for ease of use), but ultimately it is their pick, and they own their learning journey.
That means that the use of Generative AI in the class is highly idiosyncratic and specific to each student and student group.
The most frequent use cases I observed in class are seeking explanations of concepts, (unsurprisingly) working with long legal text, dealing with information architecture, rewriting the legal texts into human language, quick orientation in statutes, and of course, building the final projects.
Using Generative AI shifts the focus
With Generative AI, we can focus more on the key strategic points.
For example, if we are working with an existing legal document, just organising the information or creating section headings would take the students a lot of time.
Instead, using Generative AI, the students can extract these in seconds and then ask themselves the hard questions – does it make logical sense? Is it usable and useful? What do we need to change?
Generative AI is my Teacher’s Assistant
Using Generative AI, the students have access to their own tutor nonstop (even outside the class or when I am talking to another group).
What is more, they get valuable practice in formulating relevant questions and prompting. Plus get only what is directly relevant to their work. This is learned the hard way – it can get quite frustrating when the outcome is useless when the time is limited!
In practice, this assistance can look as providing clarity on:
- Requirements (what is a service map? What are the key components?)
- Samples – we are proposing [this service]. Prepare a service map
- Information clustering as a base for legal visualisations – here is a text, divide it into sections
- Accessibility review of legal texts – would a 12-year-old understand this?
- and a lot more that I do not know about because the students work independently and have their ingenious ways of problem solving.
And if it does not work or hallucinates? Great! It means that the students can tweak their approach and reflect on their practice and the limitations of the technology.
Some caveats to the above
This is a work in progress
When I started teaching the class, ChatGPT was published at the end of the semester, and we did only basic prompting around study guides.
As of the last semester, several students have their own subscriptions to the most powerful models and seamlessly integrate them into their workflows.
Just like the world changes a little bit with every demo day, so does my approach.
I am also fortunate that the students give me heaps of feedback to work with. If you would like to find out more about how I use data to make the class tailored to the learning needs of my students, I wrote about it here.
The educator needs to be up to speed
To do this, I spend a lot of time educating myself on these topics. I am passionate about it and I take it very seriously. But I do have this kind of space because I teach only one class consistently – this kind of prototyping and tweaking would be much harder if I had a bunch of courses to design.
Modern Lawyers is a small sample
While I am very proud of what my students achieved in the past semesters, it is important to note that this is a small class of about twenty students.
While it is great as a sandbox and a small-scale prototype, to move the needle on legal education, we would need a lot more.
The tech has its limitations
Besides the usuals, I teach the class in Czech, which is not the strongest language for most LLMs. While most of my students report being very comfortable in English, this is not something to rely on. Although I do believe English to be a key skill for the future, I do not wish to exclude anyone due to language limitations.
Also, premium subscriptions are expensive, which hurts even more on student budgets. This is of course developing with time (for example with the release of GPT4o for free users).
For example, using GPTs was just rolled out as a free feature. I do have a GPT TA, but I haven’t shared it with the students yet to avoid excluding anyone.
Final provisions
I think we are only at the beginning of Generative AI impact on legal education.
It has been a rollercoaster of prototyping, and I have been the most fortunate and grateful that the students have been so open-minded and eager to learn.
So as we continue this journey, this note will probably need an update soon.
How do you use Generative AI in your classrooms? At law schools and elsewhere?
Baru
