This work is licensed under a Creative Commons Attribution 4.0 International License (CC BY 4.0). Part of my #blogfirst Project.
Adapted from a keynote address given to the National Association of Women Judges in October 2025
At a Glance
- The Problem: The self-represented litigant (SRL) crisis is not a simple “information gap” but a “wicked problem” requiring bandwidth augmentation to help overwhelmed people navigate an intimidating system while managing personal crises.
- The Solution: AI tools can handle Procedural Work (formatting, form completion, plain language translation), freeing judges and staff to focus on Substantive Work and human connection.
- The Proof: Real courts are implementing AI solutions now with low or no budgets: Nevada’s multilingual chatbot, Michigan’s AI-assisted form completion, Florida’s SANDI navigation tool, and law school clinics partnering with courts nationwide.
- The Action: Start Monday: identify your court’s biggest procedural pain point, explore existing state resources, and pick one form that causes the most corrections and delays.
The Shared Reality
If I were to ask a room full of judges to raise their hands if they had a self-represented litigant (SRL) in their courtroom this week who was clearly overwhelmed, every hand would go up. If I asked how many have court staff who are frustrated because they spend half their day answering the same questions over and over, the result would be the same.
We all know the maze. We see it every day. For the public, it is a confusing, intimidating journey. But for judges, court administrators, and legal professionals, it is a trap, too. It is a system built on old habits and legacy processes. It is the face of institutional inertia.
We are currently in a moment of extreme hype regarding Artificial Intelligence. It feels big, fast, and urgent. But for me, it also feels familiar.
I’ve Seen This Movie Before
For the past three decades, I’ve worked at the intersection of law and technology—first as a private practice lawyer in a large firm, then as in-house counsel at Mastercard, and most recently as Director of Michigan State University’s Center for Law, Technology & Innovation where I taught MSU’s AI and the Law class. Now, I run my own small research institute, the Kennedy Idea Propulsion Laboratory, and teach a class called Legal Technology Literacy and Leadership at the University of Michigan Law School.
Throughout that journey, I’ve had a front-row seat to the evolution of legal technology. I still vividly remember the conversations I had when I launched one of the first lawyer websites back in 1995. The most common question I got from skeptical colleagues was, “Why would a lawyer ever need the internet?”
When I started one of the first legal blogs in 2003, people said, “Who has time to write for free?” And when my co-host Tom Mighell and I started our podcast in 2006, the question was, of course, “What’s a podcast?”
Each wave of technology has been met with this mix of excitement, confusion, and a healthy dose of skepticism. The AI moment we’re in today is louder, certainly, but the core challenge for us as leaders in the justice system remains the same: How do we thoughtfully and strategically adapt to better serve the public?
The question I want us to grapple with is not “How do I use AI?” Vendors are happy to answer that for you. The question is more profound and essential: How do we use AI in ways that strengthen, rather than undermine, the principles that make our justice system legitimate?
To answer that, we have to stop looking at the SRL crisis as a “kind” problem and start treating it like a “wicked” one.
Part 1: The Holmesian Reframe
The “Kind” vs. “Wicked” Distinction
In design theory, there is a critical distinction between two types of problems. A “Kind Problem” is a puzzle. It has a clear definition and a clear solution. If you are missing a piece of the puzzle, you find it, plug it in, and the problem is solved.
For decades, we have treated the SRL crisis like a “Kind Problem.” The logic goes like this: These litigants are failing because they lack information. Therefore, if we just provide the missing piece, they will solve the puzzle. We think in terms of more complete forms, more brochures, and denser FAQs.
But as every judge knows from firsthand experience, it doesn’t work. Giving a panicked, overwhelmed person a densely detailed instruction packet is like handing a non-swimmer a physics textbook on fluid dynamics. You’ve given them more information, but you haven’t solved their problem.
This is where we must take a lesson from Sherlock Holmes. In The Man with the Twisted Lip, the police saw a simple puzzle: a missing man, blood on the window. Therefore, they quickly concluded, it must be murder. All their standard tools failed because they were solving the wrong problem. Holmes succeeded because he recognized it was a “Wicked Problem”—a messy, entangled mystery involving a double life, economics, and social pressure.
As Holmes tells Watson at his moment of realization: “I confess that I have been as blind as a mole, but it is better to learn wisdom late than never to learn it at all.”
The SRL challenge is our Man with the Twisted Lip. It is not just an information gap; it is a wicked entanglement of emotional distress, cognitive overload, power imbalance, and unique personal circumstances. Simply adding more information doesn’t solve for fear, overload, or distrust.
The Bandwidth Hypothesis: Meet Emily
To understand why the “information gap” theory fails, we have to look at the human reality.
Meet Emily. She is a single mother of two, working an hourly job. Her life is already complicated. Then, her world is turned upside down. The father of her child has been arrested. She is scheduled to drop her child off for unsupervised visitation in two days. The father has already lost his temper and screamed at her over the phone. She is scared, she is overwhelmed, and she has to navigate the family court system alone.
To truly understand what Emily is going through, I want you to imagine trying to fill out a tax form for a country whose language you don’t speak or read, knowing that a single mistake could mean you get sent home or fined or worse. That is the level of stress Emily faces. Her mental and emotional bandwidth is already 100% consumed by the crisis that brought her to court.
The Vacuum of Authority
Where does she turn? She will be reluctant to call a lawyer she might have used in the past because of fear of fees. She might call a friend who is not a lawyer. She might turn to Google or generic AI chatbots. And she gets a mix of good, bad, and outright fabricated information, like an AI response that uses the custody rules for another state.
The Result: We have left a vacuum where there is no clear, authoritative guidance, and unreliable sources are filling it. The result is overwhelmed clerks, endless lines, and stacks of incorrect paperwork.
But I believe a wiser, more compassionate interpretation is this: The SRLs are not being difficult. They are intimidated and confused. Our court staff are not being unhelpful. They are systematically burned out. This is a system-wide failure, and it hurts everyone.
This is what terrible customer service looks like in our justice system. And it happens not because people don’t care, but because the court system wasn’t designed with the user experience in mind.
The Holmesian Reframe
We need a new hypothesis: Emily has a bandwidth gap.
This simple reframe changes everything. It allows us to see AI not as a fancier information provider that dumps even more information on SRLs, but as a new category of “wicked tools” designed specifically to augment SRL bandwidth and build confidence.
Part 2: The Art of the Possible – Wicked Tools
The Procedural vs. Substantive Distinction
When I say “AI,” many people imagine a robot judge. I want to be very clear: That is NOT what I am talking about.
The secret to successful AI adoption in courts is to rigorously distinguish between Procedural Work and Substantive Work.
- Substantive Work is the domain of the judge: Analyzing credibility, weighing evidence, applying law to facts, and exercising judicial discretion. We’re simply not ready for AI here and neither is AI.
- Procedural Work is the domain of the machine: Formatting, summarizing, scheduling, checking for completeness, and translating jargon.
Our goal is to use AI to handle the procedural burden so that the humans are free to focus on the substantive work that will help SRLs like Emily. We want an Infinitely Patient Clerk.
Here is what that looks like in practice for the three key players in the system.
1. Augmenting the Judge: From Archaeology to Analysis
Consider a scenario familiar to any judge: It’s Thursday afternoon. You have a motion hearing on Friday morning. The file contains a 10-page, handwritten motion from an SRL. It’s dense, emotional, and the key legal points are buried under pages of narrative not relevant to the legal issue at hand.
- The Old Way: You spend an hour deciphering handwriting, trying to separate the legal signal from the emotional noise.
- The AI Way: You feed the document to a secure AI tool. Thirty seconds later, you get a clean, one-page judicial briefing.
It gives you:
- The phonetic spelling of the names.
- A concise case history.
- The key legal issues raised.
- The specific relief being sought.
Notice what this does. It doesn’t decide the case. It transforms your prep time from archaeology (digging for facts) to analysis (weighing the facts). The AI has augmented your bandwidth, freeing you to focus on the work only you can do.
Think about the improvement in Emily’s customer experience with the court when she appears and you know how to pronounce her name, are familiar with the facts and issues, and show that you know exactly what the hearing is about.
Institutional Memory: We can also use AI for knowledge sharing. Imagine a new judge joining the bench. Instead of reinventing the wheel, they use a tool trained on the court’s anonymized orders. They ask: “What is the standard phrasing for a temporary custody order in a case involving substance abuse allegations?” The AI provides three common examples from the last five years. This promotes consistency and helps new judges become effective on day one. It also promotes consistent approach across all the judges in the court.
Document Analysis: New platforms can instantly analyze a brief to verify every citation and track every factual assertion back to the evidence. Imagine a contentious motion where one party claims the other admitted something in a deposition. Instead of your clerk spending 20 minutes searching a transcript, the AI can surface the relevant testimony, or the lack thereof, in seconds. The tool Clearbrief, which some of you might be using, does this job for you.
2. Augmenting the Staff: The End of Triage Burnout
Your clerks are currently the “human interface” for a broken system. They spend their days rejecting forms for simple errors.
- The Old Way: A clerk picks up a file, scans it, sees the financial affidavit is missing a signature, rejects it, writes a notice, and mails it. The cycle repeats three times.
- The AI Way: An AI tool scans incoming PDFs. It instantly flags the missing signature or the blank field before it even reaches the clerk’s desk. Or better yet, the filing system prevents the submission entirely until the error is fixed.
This frees your staff to handle truly complex questions or meaningfully help SRLs rather than acting as spell-checkers.
3. Augmenting Emily: The “Guided Interview”
For the SRL, the distinction between Procedural Work and Substantive Work is the difference between access and exclusion.
- The Old Way: We hand Emily a “Form.” A form is like a surprise pop quiz for a class the SRL has never taken. It asks for information in a way that requires prior knowledge of legal structure.
- The AI Way: We give Emily a “Guided Interview.”
Using AI, we can replace the form with a conversational interface that feels like a text message.
- AI: “What is your full name?”
- AI: “Do you and the other parent agree on who the child should live with?”
- AI: “Has there been any history of violence?”
Breaking a complex task into simple, manageable steps is sophisticated customer experience design. It takes a “wicked” task that requires immense cognitive bandwidth and breaks it down into tiny, manageable steps. It takes a document that consumes 100% of a person’s mental energy to decipher and transforms it into something that might only take 10%.
The Result: The Guided Interview Effect. Emily submits a perfectly completed, perfectly formatted document on the first try. She arrives at the hearing with confidence, knowing her paperwork is right. She understands what is happening because the AI translated the “legalese” into plain English for her.
She can now speak to the Substance of her case, protecting her child’s safety, because she isn’t drowning in the Procedure.
Part 3: Proof It’s Not a Fantasy
I know what you are thinking: “Show me the proof.” This is not a tech fantasy. It is happening right now in courts with real budgets and real constraints.
Courts Already Using AI for SRLs
- Michigan Legal Help: Actively developing AI-assisted form completion, plain language explainers, and enhanced online dispute resolution through the MI-Resolve platform.
- Nevada Supreme Court: Uses a multilingual chatbot to help people with court procedures, handling thousands of inquiries.
- Florida’s 11th Circuit: Uses a tool called SANDI to help people navigate the system.
- New York: The Roxanne bot helps tenants with housing repair issues, guiding them through the process step-by-step.
What’s significant isn’t just that these tools exist. Nevada is not a wealthy state with unlimited resources. They found a way to do this at scale. If Nevada can do it, why can’t you?
The Resource You Are Ignoring: Law School Clinics
I often hear, “Dennis, we have no budget and our IT department is overwhelmed.” This brings me to a resource that many judges overlook: Law School Clinics and Labs.
As someone who has taught at Michigan State University and now teaches at the University of Michigan Law School, I can tell you firsthand that today’s law students are hungry for this. They are not satisfied with just reading casebooks. They want to help build the future of the law.
You have free, expert technical assistance available to you right now through law school AI clinics and legal innovation labs.
- Stanford’s Legal Design Lab partnered with the LA Superior Court to build an AI triage tool that matches litigants with the right resources based on their specific situation in one of the largest courts in the us.
- Suffolk University’s LIT Lab is developing AI-powered court navigation tools and testing them with real users.
- University of Michigan Law School has launched an AI clinic looking for court partnerships.
- Vanderbilt, Berkeley, Georgetown, Duke. The list is growing every semester.
How a Partnership Works in Practice. Here’s how it might work for you. You identify your court’s single most problematic form. Maybe it’s a financial disclosure form where you see a 60% error rate. The clinic assigns a team of 3-4 students to build an interactive, guided version of that form as their semester project. They work with your staff to understand the requirements, build and test it, and deliver a working prototype.
- The Cost: Zero dollars.
- The Return: A custom-built tool that addresses your specific pain point.
- The Bonus: You get to see these students in action. This is the best way to identify and recruit your next law clerk and recruit someone who understands not just the law, but the technology that is reshaping it.
I can tell you this: if you reach out to any law school’s legal innovation center or clinical program and say, ‘I’m a judge interested in partnering on an AI access-to-justice project,’ you will get a response. These programs want court partnerships. The infrastructure for innovation is already built. You just need to tap into it.
Inspiration from the Private Sector: The AAA Pilot
We’re even seeing this trend in the private dispute resolution sector, which offers valuable lessons for courts. The American Arbitration Association (AAA) recently launched an AI-native arbitrator for construction cases.
The AAA chose construction cases strategically. These are high-volume, document-intensive disputes. You’ll notice that these have similar characteristics to many SRL dockets. Construction arbitration involves reviewing contracts, analyzing invoices, tracking change orders, and synthesizing technical reports. The procedural burden is immense, but much of it is pattern-matching work that AI can handle effectively.
The key lesson for courts isn’t that we should have AI judges deciding cases. Rather, it’s that AI can handle certain Procedural Work effectively when properly supervised. If the AAA, a respected institution with a 98-year history, is comfortable implementing AI in arbitration proceedings, it provides a model for how courts can thoughtfully adopt AI for procedural assistance.
The parallel to SRL cases is clear: high volumes of filings, document-heavy procedures, and parties who need accessible processes. The difference is that SRL cases have an even greater need for procedural assistance because the litigants lack legal representation entirely.
Now, this is voluntary arbitration, not court proceedings. That’s an important distinction. But it shows how these tools are entering the dispute resolution space in significant ways.
Results So Far: The Data
We are beginning to see promising data to prove this works.
- 30-50% Reduction: In pilot programs, AI-assisted intake has been shown to cut user time by 30-50%. This isn’t just efficiency for the court. It’s giving people hours of their life back. These are hours SRLs can spend preparing for the substance of their case and dealing with all the other aspects of their wicked problems.
- 90% Accuracy: AI-powered tools for determining case eligibility have matched human expert assessments about 90% of the time.
- Workload Reduction: Courts are reporting reduced staff workload, allowing clerks to spend more time on complex questions that actually require human judgment rather than repetitive form checking.
- Thousands of Inquiries: Nevada’s Supreme Court chatbot has handled thousands of inquiries in multiple languages, providing immediate answers to procedural questions 24/7. This is access to justice in action—removing the barrier of courthouse hours and language limitations.
We’re still in early days, to be sure, but cautious optimism is warranted. The data supports what common sense suggests. Automating repetitive procedural tasks frees humans to focus on work that requires judgment, empathy, and expertise.
Part 4: Your Playbook for Action
Navigating the Chaos: A Strategy for Leaders
If the tools exist and the need is there, where do we start?
The legal technology landscape today is a chaotic, overwhelming map of nearly a thousand self-identified AI tools for the legal market. As someone who built one of the first lawyer websites when you could count the legal tech companies on two hands, this is both amazing and terrifying.
Too many choices basically mean no choices. It is no wonder that you, and your IT departments, feel paralyzed. That is precisely why your approach must be to ignore this chaos around tools. Don’t start with the technology. We must resist the urge to find the “perfect” comprehensive solution and instead focus with discipline on one single, solvable problem.
The Ethical Guardrails
Judges have the highest commitment is to fairness and due process. They must be vigilant about potential risks while also recognizing that these are reasons for caution and diligence, not for inaction.
The Hallucination Problem You have heard the horror stories of lawyers submitting briefs with fake cases generated by ChatGPT. This is a real risk that gives even worse when SRLs are involved, especially when there is a lawyer on the other side. When opposing counsel cites a case that doesn’t exist, an SRL has no way to verify it. They don’t have Westlaw. This is why judicial vigilance matters. We must hold lawyers accountable when they submit AI-generated content without verification, especially in cases against SRLs. This accountability must be meaningful and not a slap on the wrist or the problem will become even worse than it is today. With SRL cases, this becomes not just an ethical issue for lawyers but a fundamental justice issue.
The Next Wave: Deepfakes. The court system is now starting to see the problems deepfakes are going to bring to evidentiary issues. I expect this is already surfacing in SRL cases: audio recordings that were never spoken, photos that never happened, documents that never existed, signatures that were never affixed. The technology to create convincing fake evidence is now accessible to anyone with a smartphone.
SRLs are uniquely vulnerable to deepfake evidence. Unlike represented parties, they don’t have access to expert witnesses who can authenticate digital evidence or forensic analysts who can detect manipulation. They may not even know to question evidence that looks legitimate. They don’t have Westlaw. They can’t check if a case is real. They’re uniquely vulnerable.
This is why judicial vigilance matters. We must hold lawyers accountable when they submit AI-generated content without verification, especially in cases against SRLs. We need clearer standards and meaningful enforcement. Courts need to be prepared to:
· Question the authenticity of digital evidence, especially in contentious SRL cases
· Require proper authentication even when evidence appears genuine
· Educate SRLs about their right to challenge evidence
· Consider appointing experts in cases where deepfake evidence is suspected
· Order sanctions and penalties on lawyers must be compelling enough to change behavior
The deepfake problem underscores why we need judges who understand AI technology and not to become technical experts, but to ask the right questions and protect due process in an AI-augmented world.
The Vendor Checklist: Red Flags to Watch For. When evaluating tools for your court, you need to be a skeptic. Here are the critical red flags that should make you walk away:
- Black Box Algorithms: If a vendor won’t explain how their tool works or refuses to provide transparency, don’t buy it.
- No Accuracy Metrics: Demand to see error rates, validation studies, and performance data.
- Vagueness on Training Data: You need to know what the AI was trained on. Does it include data from your jurisdiction?
- Resistance to Independent Auditing: The best vendors welcome scrutiny and independent testing.
- No Exit Strategy: Ensure you own your data and can export it if the vendor fails. You should never be locked into a proprietary system with no way out.
The best vendors welcome these tough questions. They understand that courts have unique responsibilities and higher standards. If a vendor is evasive when you ask about transparency, accuracy, or data ownership, walk away. There are good tools and good vendors out there. Don’t settle for anything less.
The Cycle of Innovation: Getting Unstuck
I know that shifting how we do things is hard. My own career has been defined by pivots—from tax law to technology law when the internet emerged, from private practice to in-house at Mastercard, and most recently into academia. Each change required letting go of a familiar way of working to embrace a new reality. Overcoming institutional inertia is a lot like that. It requires a conscious choice to pilot something new, even if it’s small, to see what’s possible.
This cycle is the way forward, especially when facing budget cuts and institutional resistance:
Stage 1: Partnerships
Start with those law school clinics we discussed. Reach out to the legal innovation centers at nearby law schools. Frame it clearly: “We have a specific problem with [form/process] and we’re looking for a semester project partnership.” Don’t ask them to fix your entire system. Ask them to help you solve one specific pain point.
Also consider partnerships with other courts. If a nearby court has already piloted a tool successfully, you don’t need to reinvent the wheel. Ask if you can adopt their approach or share their vendor relationship.
Stage 2: Pilot Programs
Move to small, contained pilot projects. The key characteristics of a good pilot are:
• Narrow scope: One form, one process, one specific problem
• Clear timeline: 60-90 days for most projects
• Measurable goal: “Reduce error rate from 60% to 20%”
• Reversible: If it doesn’t work, you can stop without having committed significant resources
A pilot isn’t a vague hope or a long-term commitment. It’s an experiment with a clear hypothesis: “If we implement X tool, we expect Y improvement within Z timeframe.”
Stage 3: Success Metrics
Define your metrics before you start. What will success look like? Here are examples:
• “Reduce form rejection rate from 60% to 30% within 90 days”
• “Cut average clerk time per filing from 15 minutes to 5 minutes”
• “Increase first-time-correct filing rate from 40% to 70%”
• “Reduce number of continuances due to incomplete paperwork by 50%”
Track these metrics rigorously. Create a simple spreadsheet that documents baseline performance before the pilot and tracks weekly progress during the pilot. This data becomes your most powerful argument for resources and expansion.
Stage 4: Continuous Improvement
Based on your pilot data, refine and expand. If the pilot succeeded, use the data to make the case for broader implementation or budget allocation. If it failed, document why and what you learned—that’s valuable too.
This cycle creates momentum. A successful pilot creates believers. Believers create advocates. Advocates create resources. And the cycle repeats with progressively larger and more ambitious projects.
The Monday Morning Protocol
How do we move from “Someday” to “Monday”? We resist the urge to find a comprehensive solution for the whole system. We focus with discipline on one single, solvable problem.
Here is your action plan for this week. And I mean actually this week, not “eventually.”
Step 1: Don’t Start with the AI; Start with the Pain. Take just five minutes after you finish this article to write down the single biggest, most repetitive time-waster your staff deals with regarding SRLs. Be specific. On Monday morning, ask your court administrator or senior clerk: “What is the single most repetitive, time-consuming SRL-related task you dealt with this week?” Write down their answer. That is your initial target.
Step 2: Test Drive. Visit your state’s Legal Help website. Spend 15 minutes exploring what resources already exist. Often, the tools are already there, but the court isn’t directing people to them effectively. If your state doesn’t have a robust site, visit Michigan Legal Help (https://michiganlegalhelp.org/) to see what a mature system looks like.
Step 3: Become the Pilot. Don’t wait for a committee. Pick one form. Partner with a local law school or use internal talent to create a “smart” version or a plain-language guide. Test it for 30 days. Measure the error rate before and after.
The total time investment for these three steps will be about 30 minutes on Monday morning. By Monday evening, you’ve identified your problem, surveyed existing resources, and started a conversation. You’re in motion. Motion defeats inertia.
Part 5: The Questions I’m Always Asked
When I speak on this topic, three questions come up repeatedly. They represent the biggest psychological and practical barriers to getting started. Let me address them directly.
Question 1: The IT Department Problem
“Dennis, this all sounds great, but my IT department will never go for this. They’re already overwhelmed, they’re risk-averse, and they’ll say no to anything new. How do I get past that bottleneck?”
Here’s my honest answer: You don’t need to convince your IT department first. You need to give them a success they can’t ignore.
Start with something that doesn’t require IT involvement at all. Visit Michigan Legal Help or your state’s equivalent that’s already built and hosted externally. Use it. See what works. Or use a consumer AI tool to brainstorm solutions. I give you sample results from a session I did in Appendix 1.
Then, when you’ve identified something that truly works, you go to IT with data, not a request. You say: “We tested this for 60 days. It reduced our form error rate from 60% to 20%. Here’s the data. How might we implement it?”
IT departments respond to demonstrated value, not hypothetical benefits. Show them it works first. Then they become your allies, not your obstacles.
Question 2: Where Do I Actually Start Monday Morning?
“Okay, Dennis, I’m convinced. But I’m sitting here feeling completely overwhelmed. Where do I actually start on Monday morning?”
I’ve outlined a Monday morning protocol above. Is helping improve SRL problems worth just 30 minutes of your time? As the Nike commercials urge us, just do it. If you wait for the perfect time, you’ll wait forever. The best time to start was a year ago. The second-best time is Monday morning.
Question 3: The Skeptic’s Challenge – Why This Time Is Different
“Dennis, I’ve been a judge for 20 years. I’ve seen a dozen technology initiatives come and go. E-filing was supposed to revolutionize everything. Case management systems were going to solve all our problems. Why should I believe AI is any different?”
That’s a fair question, and I respect the skepticism. Here’s why this is different:
1. The Technology Is Fundamentally Different. E-filing and case management systems were about digitizing existing processes and essentially taking paper and making it electronic. AI is about augmenting human capability. It’s not asking you to change your workflow. Instead, it’s giving you tools that make your existing workflow more efficient. E-filing didn’t make forms easier to understand, but AI can.
2. The Barrier to Entry Is Completely Different E-filing required infrastructure overhauls, system integration, and vendor contracts. It was expensive, time-consuming, and disruptive. AI tools can be tested with free, cloud-based platforms on your smartphone. You can experiment with zero institutional commitment, zero budget, and zero infrastructure changes.
Remember those kiosk projects from 10 years ago that failed? They failed because they required hardware in the lobby. Modern AI tools are mobile-friendly and accessible wherever the SRL already is: at home, on their phone, at 11 PM.
3. The Mission Is Different E-filing was primarily about internal efficiency for the court. AI for SRLs is about access to justice, a constitutional and ethical imperative. When 80% or more of family law litigants lack representation, we’re failing at our fundamental mission. This isn’t about convenience; it’s about legitimacy.
You’re Exactly Who Should Be Leading This But here’s what I really want to say to that 20-year veteran judge: You’re exactly the person who should be leading this. Your experience tells you what works and what doesn’t. Your skepticism is a feature, not a bug. It means you’ll ask the tough questions vendors need to answer.
I’m not asking you to be an early adopter who chases every shiny new tool. I’m asking you to be a thoughtful evaluator. Test it. Challenge it. Measure it. You know the problems that must be solved.
Conclusion: It Is the People’s Court
Let’s return to Emily, where we started. An Emily overwhelmed and alone, facing a system that wasn’t designed for her.
And let’s imagine an “after.” An Emily who was guided through the process step by step. Who submitted a correct form on the first try. Who walks into her hearing prepared to talk about the Substance of her case, protecting her child’s safety and her own, because she isn’t drowning in procedural confusion.
This is the Holmesian leap in action: recognizing that Emily’s struggle was never about a lack of information. It was about a bandwidth crisis compounded by an intimidating system. Like Holmes in “The Man with the Twisted Lip,” who succeeded not by gathering more evidence of murder but by seeing the wicked complexity of a double life, we must see the SRL crisis for what it truly is.
Once we make that shift from viewing this as a simple puzzle with missing pieces to understanding it as a wicked entanglement of stress, confusion, and overwhelm, we can design tools and processes that actually help. Emily’s transformation from overwhelmed and alone to prepared and confident isn’t about giving her more forms or denser instructions. It’s about augmenting her bandwidth so she can focus on what matters most to her, not the court.
When I was young, I was lucky enough to win school awards in both Mathematics and English. In many ways, I’ve spent my career trying to live at that intersection of pure logic and human language.
And isn’t that what judging is all about? You blend rigorous, analytical legal reasoning with a deep and compassionate understanding of the human story in front of you. The challenge before us with AI is exactly that. It’s not just a technical problem; it’s a human problem. It’s about using the logic of these powerful new tools to create a more compassionate, more understandable, and more human system of justice.
Emily’s transformation from overwhelmed and alone to prepared and confident isn’t about giving her more forms. It’s about augmenting her bandwidth.
This is what good customer service design looks like when applied to courts.
We must treat “Access to Justice” not just as a legal concept, but as a customer service mandate.
Because as the former Chief Justice of the Michigan Supreme Court, Bridget McCormack, always said: “It is the people’s court after all.”.
Appendix 1: The “Swiss Army Knife” in Action
A Transcript of a Live Brainstorming Session with Google Gemini Pro, edited for length
To demonstrate that you don’t need a massive budget or an IT department to start generating solutions, I acted as a Judge and ran a simple prompt through Google Gemini (a consumer AI tool).
My Prompt:
“I’m a family court judge dealing with high volumes of self-represented litigants in family law cases (e.g., custody/child support/divorce). My biggest challenge is the vast majority of SRLs do not file accurate and complete forms, generating large amounts of extra work for my staff and me and making each hearing and step in the process longer and more cumbersome for everyone than it should be. Assume that you are a world-class expert in the application of AI to assist courts with SRL issues. What are 5 practical, low-cost ways I could address this issue, particularly using technology or process improvements? Consider solutions that could be implemented within 3-6 months with a very limited budget.”
The AI Response (Excerpts):
Here are five practical, low-cost ways to address this issue within a 3- to 6-month timeframe.
1. Interactive “Smart” Forms and Checklists
Instead of providing static PDF forms, convert your most problematic forms into simple, web-based interactive wizards.
How it Works: Create a Q&A interface that uses conditional logic to show/hide fields (e.g., hiding custody sections if no children involved).
Why it’s Effective: It drastically reduces errors by ensuring all required fields are filled out.
Cost: Very low—many platforms have free tiers available.
2. Generative AI-Powered Plain-Language Resources
Use a large language model (LLM) as an internal tool to create a library of ultra-clear resources.
How it Works: Staff prompts an LLM: “Explain ‘service of process’ in simple terms for someone with no legal background.” A clerk reviews the output for accuracy.
Cost: Minimal (Staff time only).
3. QR Codes on Forms Linking to Explainer Videos
Bridge the gap between paper forms and digital help.
How it Works: Place a QR code next to the financial affidavit section. Scanning it plays a 2-minute video of a clerk explaining common mistakes.
Cost: Virtually zero (Free QR generators + YouTube).
4. Automated Filing Completeness Check
Implement a simple automated email/SMS system.
How it Works: When an SRL files, they get an auto-response: “Top 3 Things to Check: Did you sign? Did you attach financials?”
Why it’s Effective: Immediate feedback allows SRLs to correct errors before staff review.
5. AI-Assisted Document Triage (Internal)
Use simple AI to spot incomplete forms.
How it Works: Use existing PDF software to flag documents with empty “required” fields.
Why it’s Effective: Speeds up initial review; staff focuses only on flagged documents.
(I then asked the AI to rank the ideas based on “Bang for the Buck” for a budget-constrained court)
Ranked List of SRL Support Projects (Generated by AI)
TIER 1: FOUNDATIONAL “MUST-DO’S”
1. Data-Driven “Error Hotspot” Analysis (Diagnosis & Strategy) – Cost: Free
2. “Gold Standard” Exemplar Library (Clarity) – Cost: Free
3. Generative AI-Powered Plain-Language FAQs (Comprehension) – Cost: Free
TIER 2: HIGH-IMPACT INTERVENTIONS
4. Interactive “Smart” Forms (Error Prevention) – Cost: Low
5. Virtual Clerk “Office Hours” (Human Support) – Cost: Free
TIER 3: VALUABLE ENHANCEMENTS
6. QR Codes on Forms (On-Demand Guidance) – Cost: Free
7. Simple Online Submission Portal (Process Improvement) – Cost: Low
This demonstration shows that with a well-crafted prompt, even publicly available AI tools can serve as a strategic brainstorming partner, helping you generate and prioritize solutions tailored to your court’s specific challenges.
Appendix 2: Frequently Asked Questions
Category 1: Practical & Implementation Questions
Q: Our court has literally zero budget for new technology. How can we realistically do any of this? A: That’s the most important question, and it’s why this entire approach is built around a ‘no-budget’ start. For 30 days, track every rejected filing and why. This creates a data-driven list of your biggest problems at zero cost. This ‘Error Hotspot Analysis’ is free. It’s just tracking data for 30 days. The ‘gold standard’ exemplar forms cost nothing but clerk time. For technology, partner with a law school clinic. They provide technical expertise and student person-power at no cost. It’s a true win-win.
Q: My clerks are already at their breaking point. How do I get their buy-in? A: Start with their pain. Ask your most experienced clerks: ‘What is the most soul-crushing, repetitive task you do every day?’ Frame the pilot as an experiment to eliminate that specific task. Your clerks become the solution designers, not targets of change. Make it clear this is a 60-day pilot. If it doesn’t reduce workload, you’ll stop.
Q: If I do only ONE thing, what should it be? A: Do the ‘Error Hotspot Analysis.’ For 30 days, track every rejected filing and why. This creates a data-driven list of your biggest problems at zero cost. It transforms ‘forms are confusing’ into concrete problems like ‘62% of financial affidavits missing signatures’.
Category 2: Ethical & Philosophical Questions
Q: What about liability when AI gives incorrect information? A: Critical distinction: these tools provide procedural information, not legal advice. They help people understand what form to file—like a perfect paralegal, not a lawyer. Mitigate risk by keeping the human in the loop. The litigant signs and files, we just improve document quality. Courts aren’t liable when someone misreads a form packet and the same principle applies.
Q: How do we ensure data privacy and security with third-party AI tools? A: This is non-negotiable. Start with projects touching no confidential data. For tools handling forms, demand SOC 2 compliance, encryption, and contractual guarantees that you control the data.
Q: Are we replacing human connection with cold technology when litigants need empathy? A: The opposite. Technology handles repetitive tasks, freeing humans for high-value empathetic connection. Does a litigant feel connection when their form is rejected by mail for the third time? No. But when AI handles form checking and clerks offer Virtual Office Hours for personalized help, that’s real human connection.
Q: How do we prevent bias against certain populations? A: Make accessibility a day-one requirement, not an afterthought. Choose mobile-friendly tools (most SRLs have smartphones, not computers). Prioritize multilingual capabilities. If you serve Spanish speakers, support Spanish from the start. AI is getting much better at translation capabilities. Ensure screen reader compatibility for disabilities. A plain language initiative fights bias. jargon disproportionately disenfranchises lower education, non-native speakers, and cognitive disabilities. University clinics often build equity into design from the beginning. Test with diverse SRLs before full deployment. Validate, don’t assume.
Category 3: Strategic & Skeptical Questions
Q: This seems great for simple cases. Will it work for complex matters? A: Don’t think in terms of using these tools for handling trial complexity. They fix front-end intake problems that bog down all cases. Even complex cases start with paperwork. Getting that form right on the first try in 90% of cases instead of 40% helps both simple and complex matters.
Q: What is the next evolution in 3-5 years? A: In the medium term (3-5 years), expect AI-powered Online Dispute Resolution that analyzes positions, identifies agreement points, and drafts settlement language before parties reach court. For example, in debt cases, AI could identify that parties agree on the amount but not payment terms, and present options for agreement. Further out, AI will help judges with research and drafting first drafts of orders based on court standards, always with the judge remaining the decision-maker.
ABOUT THE AUTHOR
Dennis Kennedy (https://www.linkedin.com/in/dennismkennedy) is an Innovation Architect who builds AI-Human Protocols at the intersection of artificial intelligence, law, and systems thinking. His work is synthesis: connecting disparate fields to design new protocols for how humans and AI work together.
Kennedy is an Adjunct Professor at the University of Michigan Law School, where he teaches Legal Technology Literacy and Leadership. He is also Professor Emeritus at Michigan State University and Director Emeritus of the MSU Center for Law, Technology & Innovation, where he taught, among other things, AI and the Law. He is founder and principal of the Kennedy Idea Propulsion Laboratory (KIPL), a research institute focused on developing methodologies for complex, novel challenges.
Like Sherlock Holmes pursuing the “three-pipe challenge,” Kennedy focuses on problem-framing and protocol design rather than long-term implementation. His approach applies the Holmesian method to contemporary challenges: recognizing when conventional solutions are solving the wrong problem, then designing new frameworks that address the true nature of the challenge.
Kennedy is recognized as a legal technology pioneer, having launched one of the first lawyer websites (1995 – https://www.denniskennedy.com), one of the first legal blogs (2003), and co-hosting “The Kennedy-Mighell Report,” the longest-running legal technology podcast, since 2006 with Tom Mighell.
His work on AI-Human Protocols can be seen in his recent articles and heard in the discussions on his podcast. He advises a highly select group of clients on their most complex, novel challenges.
Kennedy can be reached at dmk@denniskennedy.com.
LICENSE
This work is licensed under a Creative Commons Attribution 4.0 International License (CC BY 4.0).
You are free to:
• Share — copy and redistribute the material in any medium or format for any purpose, including commercially
• Adapt — remix, transform, and build upon the material for any purpose, including commercially
Under the following terms:
• Attribution — You must give appropriate credit to Dennis Kennedy, provide a link to the license, and indicate if changes were made. You may do so in any reasonable manner, but not in any way that suggests the licensor endorses you or your use.
The licensor cannot revoke these freedoms as long as you follow the license terms.
To view the full license terms, visit: https://creativecommons.org/licenses/by/4.0/ A NOTE ON USE AND ADAPTATION
This article is released under a permissive Creative Commons license to encourage maximum adoption and adaptation by the legal community. My goal is simple: to help courts and legal professionals thoughtfully implement AI tools that improve access to justice for self-represented litigants.
You are free to:
· Use this framework in training programs or educational materials
· Adapt these principles for your specific jurisdiction or court system
· Build commercial products or services based on these methods
· Share it widely within your professional networks
The only requirement is attribution. If this work helps you serve self-represented litigants better, then it has served its purpose.
If you’re implementing these approaches or have questions about AI-Human Protocols for your court system, I’m interested in hearing about your experience.
The SRL crisis requires collective action. This work is my contribution to that mission.
– Dennis Kennedy
A big thank you to the National Association of Women Judges, Judge Michelle Rick, and Laurie Denham for giving me the opportunity to develop and present this material.
[Originally posted on DennisKennedy.Blog (https://www.denniskennedy.com/blog/)]
Want to schedule a Zoom call to talk with me about Legal Innovation as a Service, Speaking, or other services? Schedule a Zoom with Dennis via Calendly.
DennisKennedy.com is the home of the Kennedy Idea Propulsion Laboratory
Like this post? Buy me a coffee
DennisKennedy.Blog is now part of the LexBlog network.
Need a little help with your legal innovation efforts? Check out my Legal Innovation as a Service offerings.
As an Amazon Associate I earn from qualifying purchases.