In a recent New York Times article, Ann Carrns covers a report from Pew Charitable Trusts about the recent rise of debt lawsuits in U.S. courts. The main conclusion of the report is that AI makes it cheaper for debt collection agencies to take consumers to court - and thus impact them with negative judgments. I don’t disagree. But I wanted to offer a contrarian and more nuanced opinion: litigation is good and more lawsuits should be filed.

Before I get into that, let me start with a disclaimer: I do in fact run a debt collection agency. Not one going after consumer medical or credit card debt. Rather, Delos helps businesses recover unpaid invoices. Is that different? Significantly. And so are our motivations. But as far as U.S. federal and state law is concerned, we fall under the definition of a debt collection agency and must be licensed accordingly. So, I’m not going to dispute it here. In fact, my company is contributing to, and will continue to contribute to, the tally of cases filed in small claims courts. We will do so with AI. In some cases, we are, or will be, suing individuals. So, you could say I’m part of the problem.

Litigation is good and we should have more of it, even if it’s not working

Yes, dispute resolution is broken. The access to justice gap has never been wider. I’ll be the first one to say this. However, at its very heart, going to court and receiving a judgment is good. Why? Because often it’s the only way to enforce your rights. If a business contractor decided not to pay your invoice and does not respond to your messages - going to court is the only recourse you have left. I’ll go even further and say that going to court and receiving a judgment is in fact the best way to enforce your rights. A court judgment has gravitas. It communicates trustworthiness and commands respect. In fact, psychologically it seems there is nothing as desirable in a dispute as an outside arbiter resolving it - regardless of the outcome. This is what researchers describe as procedural justice: the idea that people are more willing to accept an adverse decision, comply with it voluntarily, and regard it as legitimate when the decision comes from an authority they regard as legitimate. Even when they dislike the result. There are some important caveats - like whether courts and proceedings are seen as legitimate. But all in all, courts are the most legitimate institution for resolving a dispute we have today.

A court judgment is better than the alternative. In the context of debt collection, the alternative is robocalls, nagging, aggressive demand letters, emails saying the matter will be sent to “legal” if you don’t pay and more. The Fair Debt Collection Practices Act limits many such practices, but they still persist in some shape or form, as debt collectors find innovative ways of circumventing rules. The reason such practices exist is that litigation is expensive. It’s costly for a debt collector to pay lawyers, pay legal fees and take a case to court. In many smaller-value cases, it’s simply not economically feasible. If you look at financial statements from Encore or PRA, the country’s top debt collectors, the math shows that it’s much cheaper for them to rely on call centers, negotiated payment plans and digital outreach than litigation.

But litigation today undeniably has a problem. Many problems, in fact. I won’t go into covering them here, as I’ve done so many times before, but I wholeheartedly agree with the issues listed by the authors of the report: unequal representation, procedural complexity, inadequate notice and service, insufficient scrutiny of default judgments, weak pleading and evidence, failure to enforce procedural safeguards, harsh post-judgment remedies, repeat-player advantages, and tension between efficiency and due process.

AI, more AI, always AI

I’m not a fan of castigating anyone, even debt collectors, for using AI to file more lawsuits. If the intentions behind such criticism are good (i.e., protecting consumers), a freeze might affect everyone using AI to file lawsuits. And if AI benefits debt collectors, it also benefits consumers. A recent Financial Times piece covered how U.K. courts are experiencing an increase in AI-enabled lawsuits filed by pro se plaintiffs. Granted, ChatGPT can help you only so much and a lot of work is still required, but the fact is there’s never been an easier time for self-represented individuals to file a lawsuit in court.

The French revolutionary, Georges Jacques Danton famously said “Audacity, more audacity, always audacity” to urge the French public to stand up against the enemies of the revolution. Similarly, I’m here to tell you that when it comes to access to justice, improving courts, and dispute resolution for all, we need “AI, more AI, always AI.” But not AI in the form of AI tools for judges and law clerks. I believe we need a fundamental rethinking of what civil procedure is and how it works. In fact, we need a new civil procedure paradigm (wow, big words!) - one that puts AI (and technology more broadly) at its heart. A paradigm that takes the concepts we have always associated with a rightful legal system - e.g. due process, access to justice, equity - and considers how they can be achieved with AI first and AI alone. When I say, “AI alone,” I mean AI making the actual legal decision and resolving the legal dispute. Not a human using an AI system to speed up their work. I believe such a future is: (a) necessary; (b) possible; (c) better.

It’s necessary because the number of lawsuits is about to grow exponentially. If the New York Times article mentions a deluge of such lawsuits, I’m here to tell you we’re about to face a tsunami. Most debt collection agencies I talk to are not even close to writing a lawsuit with AI. And if courts are overloaded today, then they’re certainly not ready for the time when everyone starts filing lawsuits with AI. And so, the process whereby a human reviews each case, gives it appropriate consideration, considers all facts equally, and has time to hear all parties is quickly disappearing if it hasn’t disappeared already. And a human working with an AI system is not a solution. First, waiting for an AI to do the work and then reviewing/approving it isn’t faster. At least not by the orders of magnitude we will soon need it to be. Second, such a “review and approve” process usually amounts to nothing more than rubber-stamping. And so, if the solution to courts being overloaded is a human rubber-stamping an AI’s outputs, I’d rather have an AI alone.

Now when I talk about an AI system that will autonomously rule on cases, I don’t mean the technology exists today. I’m talking about a future AI system that is capable of such a feat, but that we haven’t built yet. We shouldn’t blindly implement the latest AI models to do it. This also explains my reluctance to humans using AI to issue verdicts. A good human and a bad AI make for poor bedfellows. Nevertheless, the progress AI models have made in the last four years leads me to believe that a future where disputes are resolved solely by a machine is possible and closer than we think.

Finally, I believe this future will be better. When we think of automated AI rulings, we’re reluctant to accept them because we think a machine will be blunt, unempathetic and will lack the human touch. I believe there are two biases that fuel such beliefs. For one, we believe we hold a monopoly on empathy. We don’t. Recent research has dispelled that notion. Second, if we think of an AI judge, we think of speed, efficiency, and cost savings. Instead, we should think of it in terms of abundance. We should think of what a futuristic ideal courthouse looks like and use AI to enable its existence. If consumers today don’t understand why they’re being sued - why not have a clerk call them and explain in human terms what this lawsuit is about. Have him/her answer all their questions. Spend an hour or two if needed. Can courts afford to have humans do this? They can’t. Could an AI do it? It certainly could. It could talk to any consumer, for as long as that consumer wanted or needed, and explain everything without a single word of legalese. But let’s go further - why not make that phone conversation the consumer’s response to a pleading or summons? Why should a consumer have to file a written statement on a particular court form filled with words they do not understand (either because they don’t speak good English or because the words are legalese)? The forms were meant to facilitate the job of the court, not of the people. But with AI available, why not make anyone submit their arguments in any form and language they like? Why not let them call the court? Why not help them find the right facts? Do they seem to remember that their contract had some exception they can’t find now? Why not find it for them? What if a consumer is willing to pay but doesn’t have money? Shouldn’t the court spend time on understanding their specific circumstances, with all their details, and then making an appropriate ruling? Not only a ruling in form (“pay X”), but one in substance (“pay Y every 2 months until Z with the following exceptions”). What if a lawyered-up party writes a 200-page petition lacking substance but requiring the court to read it and the opposing party to review and respond? Today this means hours of reading time for everyone involved. Why not slash that to seconds?

The future I’m proposing is not an easy one to design, and many elements should be carefully weighed. But design it we should. And if it takes a pack of AI-native debt collectors hounding consumers in small claims courts to force us to get there, then perhaps that is what it takes.