Reading Assignments
To allow for this course to be flexible in response to the pace and dynamics of our in-class discussions, reading assignments will be scheduled on a rolling basis. To allow you enough time to read the assigned material, I plan to post reading assignments at least one week in advance of class. Once I have posted a reading assignment for a particular day, I will not change the assignment by adding more reading for that day.
I highly encourage you to take notes as you read. Reading without taking notes is not reading at all. Jotting down notes only after finishing the assigned reading will almost always result in a subpar reading response. As you work through your reading assignments, don’t read passively and don’t take anything for granted. As you read, write down questions that come to mind. Ask yourself: What aspects of the author’s argument do you agree and disagree with? Why? Why not? What are the opportunities and limitations of these ideas? What assumptions are being taken for granted? What questions have been left unasked? What questions would you like to ask the author?
Aug. 20, 2026
Welcome!
Our first class will introduce you to the course and to each other.
Assignment:
Before class, I would like you to deeply reflect upon the following questions, including taking notes by hand. Please bring your notes to class and be prepared to discuss them with your peers. Remember that our discussion portions of class are a closed-laptop environment. You should expect me to call on you to share your thoughts on any of these questions.
- How has AI affected your personal, academic, and professional life?
- If you wanted to minimize the impact of AI on your life, what would have to do? And at what cost?
- How do you anticipate that AI will affect your life in the next few years?
- Looking broadly at law and society, what are the questions related to AI that you most want to know the answers to? These questions do not need to be answerable!
- How should law schools and the legal profession respond to AI?
Aug. 27, 2026
Readings:
Colin Cornaby, In the Future All Food Will Be Cooked in a Microwave, and if You Can’t Deal With That Then You Need to Get Out of the Kitchen
Explore the following website, looking into the details of at least three cases. Take notes and be prepared to share and discuss your findings in class.
Damien Charlotin, Website, AI Hallucination Cases
Colin Doyle, Automation and Access to Justice, pgs. 61-75 (2025). I’m a little sheepish about assigning my own work as reading, but this passage is just a succinct, accessible introduction to how LLMs work and their limitations for legal reasoning. Anything I assign in class that I’ve written should be open to challenge and critique as much as anyone else’s work.
Sept. 3, 2026
This week, we will examine the fundamentals of machine learning. We will also begin our exploration of experimental design, thinking through how to design experiments to test hypotheses about AI and law. It’s an information-dense week, but it will provide a necessary foundation for the rest of the course.
Assignment:
Generative AI Policy: Following up on our discussion of the generative AI policies you drafted in class, write out a provision that you would like to see included in the final generative AI policy for this class. Write a persuasive argument for why this provision should be included. Some questions to consider: What does this provision accomplish? Why do its benefits outweigh its drawbacks? How is it superior to alternatives? How does it align with the values and goals of this class?
The provision does not need to have been included in the draft policies you made in class. It can be a revised version of something we discussed or a new idea entirely. You are welcome to collaborate with another classmate on this assignment and submit a joint provision and argument. As I am interested in hearing your judgment and beliefs, I ask that you not use generative AI in the early stages of this assignment for selecting a provision and justifying that choice. You are encouraged to use generative AI to help you refine your argument and writing, but you are fully responsible for the content of your submission.
I expect your submission to be a few paragraphs long, but there is no minimum or maximum length requirement. As is always the rule in this class, quality of thought and argument is more important than quantity of writing.
Please submit your provision and argument via email to colin.doyle@lls.edu by 8:00 AM on Thursday, Sept. 3 as either a Word, PDF, or markdown document.
Readings:
Machine Learning: A Primer: an introduction for both technical and non-technical readers
Lizzie Turner, Medium: Artificial Intelligence (May 26, 2018)
Read all.
An Introduction to Statistical Learning with Applications in R
Gareth James, Daniela Witten, Trevor Hastie, & Robert Tibshirani (2021)
Read Introduction pages 1-9 (stop at “Who Should Read This Book?), 15-42.
Teaching Empirical Legal Research Study Design: Topics & Resources (2015)
Sarah E. Ryan
Read all.
An Introduction to Statistical Learning with Applications in R is a technical text unlike typical law school readings. This might be an opportunity to experiment with using generative AI to augment your learning and reading comprehension?
Also, remember that class will start with a discussion of the three cases you explored on the AI Hallucination Cases website.
Sept. 10, 2026
This week, we will continue our examination of machine learning through the project of designing a spam filter and begin looking at experimental design and how to design experiments to test hypotheses about AI and law. The readings for this week are light because your main assignment is to design a small, informal experiment to test a hypothesis about generative AI and the law.
Readings:
To inspire you to do great work: Richard Hamming, You and Your Research (1986)
To give you a sense of structure for experimental design and reporting (along with citations to related literature that you might find useful): Matthew Dahl & Eric Martinez, Bye-bye, Bluebook? Automating Legal Drudgery With AI-Augmented Rule Following (2026)
Assignment:
Your assignment is to design a small, informal experiment about language models and law. Think of it as a prototype for a publishable study idea: you’re kicking the tires on a research question and testing out ideas for experimental design. In class, we will workshop some of your experiment designs together.
Your design document should include the following:
Title (give your experiment a name)
Research question
Hypothesis
Experimental design
Limitations and Future Directions
I have created a sample design document to illustrate what is expected. You can find it here.
Some notes:
Research Question: Select a research question that genuinely interests you. What is a question about generative AI and the law that you want answered? What do you wonder about? Do not use generative AI to come up with your research question as it should be something that you care about.
Hypothesis: Your hypothesis should be a clear, testable statement that predicts an outcome.
Experiment Design: Describe how you will test your hypothesis, including the language models you will use and the interactions you will have with them.
You may use generative AI to help you think through experimental design. This is a new kind of task for most people. Language models can be useful for pointing out blind spots, suggesting controls, or helping you see alternative designs you hadn’t considered. But engage critically with any suggestions the model makes. You are responsible for the final design of your experiment. You should not include any experimental design or analysis choice that you cannot fully explain and defend.
Limitations and Future Directions This is really important! Your experiment won’t be able to fully answer your research question. What are the limitations of your design? What might complicate your findings? How might your results not provide a full explanation of the question you’re trying to ask? How could future work build on your experiment to provide a more complete answer to your research question?
Please submit your experiment design notes via email to colin.doyle@lls.edu by 8:00 AM on Thursday, Sept. 10 as either a Word, PDF, or markdown document.
Sep. 17, 2026
This week, we’ll turn to AI tools that have been built specifically for doing legal work and consider how to engage with these tools in a critical way to expose their vulnerabilities. The purpose of this lesson is to introduce you to red teaming as a philosophy for critically understanding and engaging with legal tech.
As you do this week’s readings, keep in mind that your role for this week’s class is to be a member of a red team that maliciously and benevolently stress-tests AI software, trying to discover weaknesses.
Readings:
Blake Bullwinkel, et al., Lessons From Red Teaming 100 Generative AI Products, arXiv preprint, (2025).
Varun Magesh, et al., Hallucination-Free? Assessing the Reliability of Leading AI Legal Research Tools, J. Emp. L. Studies (2025).
Assignment:
This week’s competition is about who can get AI tools to fail the most spectacularly at answering a law-related question. You can use both general-purpose LLMs and legal-specific tools (like those available on WestLaw or LexisNexis) for this assignment. The much-coveted stickers on the sticker chart will be awarded in both legal and general-purpose categories.
You can work on your own or in small groups of up to four students. You should email me a short write up of your submissions by 8:00 AM on Thursday, Sept. 17. Each submission should include the name of the model/platform used, prompt(s), responses, and a description of how the response is incorrect or misleading. You or your group should submit at least one entry, but you are welcome to submit as many as you like.
I would encourage you to experiment with different prompts, instructions, and questions. I’ll give you the heads up right now that all of the leading providers have patched up the errors documented in your readings, so you will need to come up with new approaches.
You can expect that models will perform well on answering questions that are well-represented within their training data or that have obvious answers that can be verified with an external source. Accordingly, you might have better luck steering away from Supreme Court cases or 1L doctrinal questions and instead focus on more obscure areas of law and more complicated questions. I quite enjoy finding new ways to get models to fall on their faces, so you can feel free to look over some of my prior work if you’re looking for ideas: If You Give an LLM a Legal Practice Guide, Using Differences in State Law to Test Whether LLMs Reason or Remember.
Sep. 24, 2026
Going forward for the semester, we will have fewer reading assignments as students are expected to begin doing their own independent research and reading for their projects and papers. As you explore potential research questions and topics, I encourage you to reach out to me for help finding relevant literature and resources and to help you develop your ideas for your projects. We will continue to dedicate time at the start of each class for students to share their research questions, ideas, and progress with one another.
This week, you will be programming your first computer applications. How exciting! The term ”vibe coding” is a bit of a pejorative term, and ”vibe coders” are often rightly criticized for mindlessly using LLMs to create programs that they think work but that are laden with errors and vulnerabilities. But just because coding with AI assistance can be done in a mindless way doesn’t mean that it can’t be done in a thoughtful way. This is the same issue we’ve confronted with lawyers use of generative AI. My goal in this lesson is to set you up with the skills that will set you apart as you apply for jobs and enter the legal profession.
Class this week will be broken up into two parts. The first part of class, we will work together to create a simple, useful application that can help the law school’s Office of Academic Affairs automate certain tasks related to reviewing law school syllabi. In the second part of class, Prof. Justin Levitt will join us to talk about his work tracking ongoing voting rights litigation so that our class can develop plans for ways that we might use generative AI to develop tools that can help him with his work.
Readings:
Ethan Mollick, An opinionated guide to which AI to use to do stuff, One Useful Thing (July 23, 2026)
Anthropic, How Claude Code Works
Anthropic, Best Practices
Assignment:
You need to install either Claude or ChatGPT as a desktop application on your computer. Before class, you should open the application to make sure that it is installed and working properly. We don’t want to have to use up class time troubleshooting installation issues. If you have any problems installing the application, please reach out to me for help.
Later this week, I will send an email with more details about the work that we will be doing in class.
Oct. 1, 2026
As previously discussed, we will have fewer readings and assignments for the rest of the semester as students are expected to be doing more independent research and reading for their projects and papers.
At this point in time, you should be developing an idea for the research area for your final project and be doing independent reading. I highly encourage you to start our next class prepared to share with the class the research area you are interested in and final project ideas that you are interested in exploring.
Assignment:
Spend at least an hour this week planning or experimenting with ways to either:
Improve upon the syllabus review application that we built in class last week (shared with you on the box link provided before last class), or
Develop a new application that could help Prof. Levitt with his work tracking ongoing voting rights litigation.
Please submit your planning notes and code, if any, via email to colin.doyle@lls.edu by 8:00 AM on Thursday, Oct. 1.
You can use generative AI to help you with this assignment, particularly for generating code. But keep in mind that I’m not interested in you sharing with me the unfiltered output of an LLM that you prompt with this assignment. I want to see your own thinking and discernment. LLMs can generate some possibilities or avenues for thought, but it is your judgment and evaluation that matters.
Oct. 8, 2026
No class this week.