Sign in Subscribe

Instant analysis of new cases with AI

Instant analysis of new cases with AI

It’s an early morning Thursday in June. The Supreme Court has just released a major opinion—one that will impact schools and children with disabilities across the country. You get some emails. One is from the university's media department who wants you to connect with a reporter from a local TV station newspaper. The second is an email from the local NPR affiliate. They want your expert take, and they want it now. For years, this scenario meant either turning down opportunities so that some other academic could get credit for a hurried review of a complex case or your own frantic skim of a 50-page opinion, a gut-check based on your expertise, and the prayer that your off-the-cuff analysis holds up. Today, we have a better way.

Generative AI is a transformative tool for legal academics and students, capable of turning hours of analytical work into minutes of strategic preparation. I want to walk you through a process I recently completed using Google's Gemini to demonstrate how you can leverage these tools to go from a dense slip opinion to clear, insightful commentary for multiple audiences in 30 minutes. The subject is today's Supreme Court decision in A.J.T. v. Osseo Area Schools, but the focus here is on the powerful, replicable workflow. By the way, while this blog entry emphasizes speed and immediacy for breaking news analysis, it's worth noting that this same methodology yields even richer and more nuanced insights when applied to established cases where time constraints don't limit the depth of research and reflection.

The Three-Step AI Workflow: Analysis to Accessibility

My goal was not simply to "summarize" today's decision in that case. It was to build a layered understanding that could be deployed for different purposes. I fed the AI a downloaded PDF of the AJT case and prompted the AI to generate three distinct outputs, each serving a unique function.

My request to the AI was specific. I didn’t just ask for a “summary.” I tasked it, acting as a specialized “Legal Case Briefer,” to produce a comprehensive, multi-part package. The first product, a traditional 1L-style brief of the case was produced by using a Gemini Gem I had previously constructed. You can retrieve that Gem and the equivalent CustomGPT here. The remainder of the prompt is set forth below.

Brief this case per your instructions. Then write a 1000 word "Nina Totenberg" or "Linda Greenhouse" style news article about it. Then give me 10 pull quotes for newspaper reporters in response to likely questions.

Within minutes, the AI delivered three distinct and valuable products:

Step 1: The Deep Dive - The Enhanced Case Brief

First, I tasked the AI with creating a traditional, 1L-style case brief, but with a significant upgrade. I asked for not just the facts, holding, and reasoning, but also for a detailed breakdown of the statutory scheme, an in-depth analysis of the concurrences, a set of five hypotheticals to test the boundaries of the new rule, a critique of the opinions, and key quotations.

  • Why it's valuable: This is the foundation. Before you can explain a case to anyone, you must understand its nuances yourself. The AI’s ability to quickly extract the core reasoning from majority and concurring opinions, and even to generate plausible hypotheticals, is a massive accelerator. The critique section forces a move from mere comprehension to evaluation, preparing you for the "so what?" questions. For a professor, this detailed brief is an instant set of teaching notes. For a student, it’s a masterclass in reading a case from all angles.

Step 2: The Public Narrative - The News Article

Next, I instructed the AI to switch hats completely and become a veteran Supreme Court reporter like Nina Totenberg or Linda Greenhouse. The task was to write a 1000-word news article explaining the A.J.T. decision to an educated but non-legal audience.

  • Why it's valuable: This is the crucial act of translation. Legal experts often struggle to shed jargon and frame a legal issue as a compelling human story. Prompting an AI to do this forces a clarification of the core narrative. What are the stakes? Who are the winners and losers? What larger conflict does this case represent? The AI-generated article serves as a perfect practice run for talking to a reporter. It helps you structure your thoughts, find your key points, and anticipate how the story will likely be framed in the media.

Step 3: The Rapid-Response Toolkit - The Pull Quotes

Finally, I asked for the most distilled output: ten "pull quotes" a journalist could easily slot into an article. These needed to be short, punchy, and accurate soundbites that addressed the case's holding, its impact on families and schools, and the ideological tensions revealed in the concurrences.

  • Why it's valuable: This is your press-ready toolkit. When a reporter calls on a tight deadline, they are often looking for a sharp, 15-second clip or a two-sentence quote. Having these prepared allows you to deliver insightful, accurate commentary without stumbling. It helps you control the narrative and ensures your main points aren't lost in a longer, more meandering conversation. It’s the difference between being a helpful source and being the go-to expert.

This multi-layered approach offers immense value to both professors and students.

For professors, the benefit is clear: it tames the chaotic news cycle. The AI doesn’t replace the professor’s expertise; it augments and accelerates it. By handling the initial, time-consuming labor of summarization and translation, the AI frees up the professor's cognitive bandwidth to focus on what truly matters: providing the high-level, nuanced critique and forward-looking analysis that only a human expert can. They can walk into a media interview not just having read the case, but having already processed it from multiple angles. This is all the more true if you ask multiple LLMs to perform similar analyses or use some other CustomGPTs I have created that are designed to provide multiple perspectives on a case or a focused CRT or conservative slant.

For law students, this is a supercharged learning tool. When a major case drops, they no longer have to wait days for a syllabus note or a commercial case summary. They can get a comprehensive first pass—the facts, the holding, the reasoning of every Justice, and potential applications—almost instantly. This allows them to engage with new law in real-time, preparing them for class discussions and helping them develop the critical skill of seeing a single legal event through multiple lenses: the doctrinal, the political, and the public narrative.

A Multi-Model "Ensemble" Strategy

For even more robust results, consider using multiple AIs in concert. You could run the same prompt on both Gemini and ChatGPT (see results here) and compare the outputs. This is an excellent way to cross-check for factual errors or "hallucinations." You might find that one model’s explanation of the procedural history is clearer, while the other’s critique of the concurring opinions is more insightful.

This "ensemble" approach allows you to become a curator of AI-generated content. Indeed, you can feed multiple results into a "master" LLM and try to extract the best elements from each model's output. You are no longer just a consumer of information; you are an editor-in-chief, directing a team of incredibly fast, knowledgeable, and tireless research assistants. Here are my results using Gemini and ChatGPT as the research assistants and Claude (with these instructions) as the editor. Truth be told, however, I prefer the Gemini response over either one from Gemini or the Claude synthesis. This is all still a bit of an art form.

For professors, this AI-driven workflow is about more than saving time. It’s about elevating the speed and quality of your public engagement. It's a tool for thinking, helping you to structure, translate, and distill complex information on demand.

For students, this process is an active learning goldmine. Instead of just reading a case brief, a student can prompt an AI to create one, then critique the AI's work. They can learn to translate legalese into plain English by generating a news article and then refining it. It teaches not only legal doctrine but also the vital meta-skill of effective legal communication.

The integration of generative AI into the legal profession is inevitable. It is not (quite yet) a replacement for human expertise but a powerful amplifier of it. By embracing these tools, we can not only keep pace with the relentless flow of information but also enhance our ability to understand, critique, and, most importantly, educate others about the law. The next time a big case drops, don't just read it—put your AI assistant to work and see how fast you can go from slip opinion to shaping the conversation.

By the way, did I actually write this article less than an hour after the Supreme Court released its opinion? Well, not exactly. I "vibe wrote" it. Meaning at the end of my colloquy with Gemini I asked it the following:

Now write a 1000 word blog post for legaled.ai, a new AI in legal education blog that I have created, describing what you just did, why it is valuable for law professors who interact with the press or even students seeking quick understanding. Explain the mechanics of how we proceeded using a "Gem," how this could equally well be done using ChatGPT, how would could even meld multiple AI answers together. The idea is to focus not so much on this particular case as on the process of using AI to quickly gain an understanding of new developments.

It then produced something that roughly reflected my intent. I then spent 30 minutes editing it. And, voila. This post.

Too busy to actually read this article. I fed it to NotebookLM and asked it to create a short podcast based on it. Here are the results. https://notebooklm.google.com/notebook/edead723-8c3b-4468-84b2-0a43c85a16f1/audio

Addendum: I just gave the New York Times account of the court's opinion ("Article A") and my AI-generated account ("Article B") to an AI (Claude 4 Sonnet) and asked which was better. Turns out that AI likes AI. "Article B is significantly better than Article A," Claude wrote. You can see its explanation of why the AI-generated account beat the New York Times here.