abstract. Access to justice in American civil courts won’t come through free or pro bono lawyers. To drive down costs, we need to loosen bar regulation and streamline procedures. And we should embrace technology and AI responsibly to give more people the legal help they need but can’t afford.

Introduction

Justice isn’t cheap. America’s legal system works for big disputes—those worth, say, fifty thousand dollars or more. Contingency-fee lawyers take those cases, and litigation finance helps pay for even bigger ones.1 But numerically, that’s only the top of the pyramid. The pyramid’s base comprises many more, smaller-dollar disputes and transactions: wills, divorces, custody of children, apartment leases, employment contracts, and the like.2 Poor and even middle-class people often can’t afford to pay two hundred dollars per hour to get divorced, negotiate away a noncompete clause, or fight an eviction.3

That rate seems high to ordinary people, yet low to lawyers. Lawyers earning two hundred dollars per hour may struggle to pay overhead: rent, secretaries, paralegals, and Lexis and Westlaw bills, not to mention student loans. Our expectation of bespoke lawyering for every need, big or small, dose’t work financially for small-dollar cases.

Lawyers like lawyers. Lawyers trust lawyers. And when you have a hammer, everything looks like a nail. So when lawyers see pro se litigants, they (we) naturally think the solution is to come up with more free lawyers for them.

Unfortunately, that approach hasn’t worked. The American bar has been trying it for decades, without success.4 No cavalry is going to charge over the hill to the rescue. And the amount of pro bono legal help available is dwarfed by the scale of unmet legal needs. Some surveys report that nearly thirty percent of American legal needs go unmet, with people “lump[ing]” it rather than pursuing any kind of redress.5 In most other countries, much of that thirty percent gets help from paraprofessionals, but we forbid that as unauthorized practice of law.6 That lack of access to justice is a crisis, and over the decades it has only grown.7

Yet the American bar cartel is not responsive. And a proper response could threaten many lawyers’ livelihoods. Many depend on getting potential clients to scrape together whatever funds they can to pay for legal services. Free or low-cost legal services on a massive scale would undermine many lawyers’ current business models.8 It’s hard to help those who can’t afford services without also helping those who stretch to pay but would much prefer to pay less. But even though reforms would hurt those lawyers, they would help many more people who can’t afford legal help.

Part I of this Essay advocates increasing access to justice by loosening bar regulation. Though deregulation would be disruptive, its benefits outweigh the costs. Bar authorities can license paraprofessional help, much as medical authorities let nurse practitioners, midwives, urgent-care clinics, and the like supplement doctors and hospitals. They can also authorize faster, cheaper paths to the bar than three years of full-time law school. Judges must play a role too. We need to streamline our procedures, experiment with inquisitorial and small-claims approaches, and be more patient and accepting of pro se litigants. Clerks of court should lend a hand too.

Loosening up bar regulation will also let technology play a bigger role, as Part II explains. Legal websites, fillable forms, and courthouse computer kiosks can help. But bigger change is at hand: artificial intelligence (AI). All we read about AI in the legal field are the massive failures—the hallucinated authorities and ideological distortions.9 Thus, some courts have forbidden using AI.10

Despite these and other downsides, AI holds enormous promise. It’s probably already better than pro se litigants are, and it certainly will be soon. Rather than trying to prevent change, courts and the bar need to loosen up and experiment with AI to help with everything from litigation to contract drafting to employment advice.

Of course, there must be checks. Courts should keep enforcing rules like Rule 11 and explain litigants’ obligations to verify everything they file. We should distinguish between general predictive AI and tools trained on reputable legal authorities. But, I conclude, we’re better off embracing the future responsibly than resisting it. Even if we can’t give everyone lawyers, we can at least give more people the legal help they need.

I. lowering bar barriers

Bar regulation is supposed to be consumer protection. By insisting on three years of full-time law school followed by a comprehensive, state-specific bar exam, bar authorities hope to guarantee top-notch representation.11 Legal education is long and costly. State-specific licensing creates more barriers to entry. And lawyers are careful about tailoring their help to each individual client and case. For the high-dollar cases, this model may work. But at the bottom, it backfires. The lucrative cases get a Cadillac approach with a Cadillac price tag, while the small ones are priced out of the market.12

Overregulation thus hurts clients with less money and smaller cases. And it’s dubious how much it helps the rest. One would expect that true consumer protection would mean suspending and disbarring inept lawyers. Yet these sanctions are rare.13 It’s hard to get into the bar and even harder to get tossed out of it. In practice, the regulations are more suited to protecting lawyers’ turf than to protecting poor clients.

The organized bar’s standard response is twofold. First, for decades, progressive lawyers have sought a civil Gideon right to appointed counsel for important civil cases, like immigration, landlord-tenant, and child custody.14 Pointing to the large numbers of people who lack the resources to hire legal counsel in civil cases, these advocates have argued that, like in criminal cases, poor litigants should be guaranteed a free, appointed lawyer.15 But this legal campaign has gone nowhere. A little more than a decade ago, in Turner v. Rogers, all nine Justices of the Supreme Court rejected the argument that due process requires giving an allegedly “deadbeat dad” a free lawyer before holding him in civil contempt.16 State courts and legislatures have likewise refused to mandate or fund a civil Gideon right.17 As I’ve argued elsewhere, it’s just not happening.18

The bar’s second response is almost as unhelpful. Bar authorities have long called for lawyers to provide pro bono legal assistance to the poor.19 But the amount of pro bono help given is dwarfed by unmet legal needs.20 And as mentioned, a massive pro bono campaign would undercut the business model of family lawyers, trusts-and-estate lawyers, employment lawyers, and the like, by giving their potential clients free legal services.

Professional protectionism is understandable. Competent legal help matters. But we’ve made the best the enemy of the good, leaving huge numbers of people without any assistance. Yes, people need good legal help. They also need healthful food and good medical care. And we don’t limit food to organic produce or fancy restaurants, nor do we limit medical care to doctors at top-flight hospitals. On the contrary, groceries and restaurants sell a wide range of food at many price points, and patients have many alternatives to doctors and hospitals, including midwives, nurse practitioners, telehealth, urgent-care centers, and minute clinics. These alternatives are often faster, cheaper, and more convenient.21

In a choice between lawyers’ entrenched interests and the public interest, the public’s access to justice should win. Bar authorities should experiment with simpler, cheaper ways to train lawyers, like apprenticeships, remote education, more part-time and night education, and law degrees focused on specialties like criminal defense or consumer law.

Over the past few years, we’ve seen how law students can learn remotely. The COVID-19 pandemic forced many law schools to transition to virtual platforms like Zoom. Although there were a few bumps along the way, schools made the switch successfully.22 Even post-pandemic, some schools continue to offer remote and hybrid courses.23 That is a step in the right direction. In-person education remains the gold standard, but remote and hybrid options can bring down costs and make legal education more flexible, especially for students juggling families and careers.24

Bars should also experiment with limited-license legal professionals, paralleling what many other countries do. Many Latin American countries, for instance, let notarios handle simpler legal issues, bridging the gap between high-cost professionals and overworked aid organizations.25 The American bar could follow their lead, letting paralegals and social workers do the same.

A few American states timidly experimented with that, but they may not have lowered the costs and training requirements enough to let these programs succeed.26 For example, Washington State lets licensed nonlawyers, called “limited license legal technicians,” advise and assist people going through “divorce, child custody, and other family law matters.”27 The tasks they handle are simple ones that do not require three years of specialized legal training. This is a step in the right direction, but the scope of what these technicians may do is still too narrow to make a serious dent in the access-to-justice problem.

True, these measures will disrupt some segments of the bar. That means some pain for existing market participants, much as the rise of Uber hurt existing taxi companies.28 But it also grows the size of the pie, as people who previously got no legal help now get some (much as the advent of Uber and Lyft seems to mean more rides total, not just the cannibalization of existing taxi trips29). That trade-off seems typical of what Joseph Schumpeter calls capitalism’s “creative destruction,” in which industries evolve by finding more cost-effective ways to deliver goods and services.30

More choices are generally better. Consumers with more money or higher-dollar legal matters are still likely to seek traditional lawyers for traditional services, much as wealthier patients still get fee-for-service medical care. Poor and middle-class consumers will reap most of the gains, as they have more options within their price ranges.

Courts should also do their part. Too many courts tell their court clerks not to say anything to pro se litigants lest they commit the unauthorized practice of law.31 They could at least point litigants in the right direction, offering them some basic tips and electronic resources. Courts could, for example, give pro se litigants informational guides that explain what documents they must file to start a lawsuit, or could help connect them with local legal-aid organizations that would be willing to help.

Courts can also streamline procedures. Currently, procedural rules are designed by lawyers for lawyers. Adversarial procedures presuppose a competent, well-funded lawyer on each side to explore and exercise the various options. But civil courts need not be adversarial. Instead of relying on a lawyer to frame the issues and conduct discovery, judges or court clerks can inquire into the basic facts and common issues and defenses. That sounds like an exotic European import. But it’s also how Social Security administrative law judges already handle disability cases.32 And it’s how family courts must gauge an allegedly absentee parent’s ability to pay before holding him or her in civil contempt for not paying child support.33 In simple cases like these, the costs of an adversarial system (namely, expensive lawyers and long, drawn-out proceedings) outweigh the benefits (such as exhaustive discovery, briefing, and argument). So the parties are better off having the court take over and expedite the proceedings.

Another approach is to expand small-claims courts. Those courts use procedures designed for pro se litigants and sometimes even ban lawyers.34 These slimmed-down procedures quickly and cheaply resolve minor disputes.35 For example, a study of a county’s small-claims court system in California found that the average time from filing a complaint to trial was fifty-seven days.36 That is far faster than the multiple years it usually takes for cases to get to trial in federal court.37 The study also found that the county’s small-claims court processed around a whopping five thousand cases in a year, using a single courtroom and with far lower administrative costs than the usual system.38

Expanding small-claims courts’ informal, efficient procedures to other minor-dispute contexts would reduce the cost and access barriers to poorer litigants. Though the procedural protections afforded in the normal court system are important in high-value and high-stakes cases, the benefits of reduced procedure seem to outweigh the costs in smaller cases. If asked to choose between resolving their legal problems with reduced procedure and “lumping it,” surely most litigants would choose the former.

Plus, there’s surprisingly little evidence that lawyers are essential to fairness. What evidence we have suggests that the answer depends on the type of procedure. In courts that use formal, adversarial procedures, having a lawyer seems to matter.39 But in courts with informal, less adversarial procedures, it’s not clear that a lawyer matters much.40 As it turns out, much of the value that lawyers add comes from their ability to navigate complex, formal procedures. So minimizing procedure in simpler cases would let poorer litigants get fair results without needing to shell out for a lawyer.

In short, the bar and courts can work together to streamline procedures and offer alternative legal help. Perhaps the most promising source of alternative help may come from technology, as the next Part explains.

II. ai and other technology

The bench and the bar are innately cautious, and with good reason. The rule of law has grown sturdy over time, like a mature tree, and procedures survive often because they have proven useful and workable. We judges are former lawyers. We are comfortable with and used to the bar cartel. We become judges after lengthy careers as members of the bar. We didn’t grow up using the most recent tech, we don’t understand it well, and we aren’t always comfortable with it. Understandably, we hesitate to change.

But as Edmund Burke teaches, we must couple our disposition to conserve with an inclination to improve.41 And modern technology offers many ways to increase access to justice. Websites like Google Scholar and Cornell’s Legal Information Institute have long made cases and statutes available to all for free.42 Fillable forms drive down the costs of basic wills, leases, and other contracts. Computer kiosks in clerk’s offices can offer step-by-step instructions and forms to guide pro se litigants.

Software can do more than just provide information and templates. Online dispute-resolution services like Modria automate mediating parties’ disputes, usually obviating the need for a human mediator.43 LegalZoom and Rocket Lawyer help incorporate and set up businesses cheaply, and, for an extra fee, they offer a lawyer to consult and personalize those standard arrangements.44 By unbundling legal services per transaction, they provide a range of affordable price points.

These technologies, though helpful, can take us only so far. They are cookie cutters for standard, high-volume matters. Options are somewhat limited. They are mostly static, not designed to learn and evolve. They automate a system designed for lawyers but (except maybe for mediation) don’t disrupt or replace it.

That’s where AI comes in. The downsides are well known. We hear horror stories about AI hallucinations and algorithmic bias.45 Some courts, including a federal court in Ohio, have reacted by banning its use.46 Poor people may be less literate and have less computer access, creating a digital divide between the powerful, expensive AI software available to top law firms, and the free AI available to the masses.47 And any major new technology will make bloopers—some hilarious, many embarrassing. The problems are apparent.

Even so, the upsides are substantial. Legal research, historically difficult and time consuming, becomes easier and more efficient. Many of the new AI-powered search tools are about as good as a second- or third-year associate, quickly drafting useful paragraphs with jurisdiction-specific citations. For instance, Westlaw’s AI-enhanced search tool lets users input natural-language legal questions. The tool then responds with not only a list of relevant cases, but also a short paragraph answer.48 Similarly, Lexis’s offering, Lexis+ AI, provides conversational responses to legal questions, summarizes areas of the law, and drafts basic memos and emails with minimal user inputs.49

Though they are still far from perfect, these sources trained specifically on legal materials are far less prone to hallucinate than generic large-language models like ChatGPT.50 These legal-specific tools already appear to be better than most pro se litigants.51 And assistants like CoCounsel promise to bring down the costs of tasks like document review and contract revision.52

Rather than reflexively opposing innovation or wildly cheering it on, courts need to guide these experiments. Rule 11 and the like already let judges sanction lawyers who fail to take reasonable steps to verify the facts, legal authorities, and arguments that they advance.53 Judges can also use their discretion to apply the same rules to pro se litigants, taking into account that these litigants can’t always tell when AI tools spit out hallucinated authorities. Those rules matter. But they should be guardrails, not barriers to prudent experiments. I’m not sure that AI-specific bans or certifications add much (if anything) to Rule 11, though of course policies can remind lawyers and litigants of their obligations to check their work.54 Indeed, courts could even use pilot projects in simple, low-stakes cases to compare motions drafted with and without AI assistance. My guess, based on the briefs I’ve seen, is that many lawyers will do better work once they learn how to use AI while checking and editing its output.

Ultimately, we must get past the dichotomy of man versus machine, as if we need to choose between John Henry and the steam-powered drill that he famously raced.55 The sweet spot is likely a hybrid model, with pro bono lawyers and, when possible, pro se litigants leveraging and guiding computer-driven legal assistance.

Capitalism’s creative destruction comes with costs.56 And now the costs of automation will fall on lawyers, the quintessential white-collar workers. Computers have already cannibalized a lot of routine document review and searching, and now they will drive down the cost of menial work like basic drafting.57 This development seems likely to play into economist Tyler Cowen’s framework that “Average Is Over”: the top fraction of legal talent will still be paid well by the richest companies on the highest-stakes cases.58 But plunging costs will democratize legal assistance for most other matters, helping poor to middle-class consumers at the expense of many lawyers. That may dent legal employment and thus hurt some law schools. Like all automation, it’s a tradeoff, but one that promises to bring down the ridiculously high cost of legal assistance, particularly for those of modest means.

What else will we lose with automation? Well, one cost may be social and emotional. Modern technology-powered medical care has proven more efficient but less humane, as people see doctors and nurses for shorter time slots, sometimes by Zoom or phone, or not at all when they get help from websites and apps.59 Some patients doubtless miss the human connection of chatting with their personal doctors and getting to know them over time. Interacting through a screen or phone is not the same as doing it in person. The same is true of lawyers. But let’s not romanticize the alternative. It’s not as if poor people on long waiting lists for legal aid have time to share their case’s details and personal dimensions with the lawyers whom they have difficulty securing. In practice, some help is better than no help at all.

That social and emotional loss suggests where to limit automation. As we triage which roles demand personal counseling, we should keep human lawyers actively involved in prosecuting and defending felonies and maybe in divorce and child-custody cases. Those clients are understandably distressed, despondent, and even suicidal. Emotions and grievances loom large in negotiating family-law resolutions. AI assistance can still help those human lawyers improve their representation. But human emotional support and moral counseling is the last thing we should farm out. Lawyers are counselors, not just brains in jars (or silicon).

Like many judges, I have mixed feelings about automating the law. At their best, lawyers offer their clients wise counsel and empathy, humane goods that no robot can supplant. If I were charged with a serious crime or had to fight for custody of my children, I’d want the best, wisest, most humane lawyer money can buy. But that’s an ideal. Many lawyers fall short of it; they may be too busy, out of their depth, or simply not very good. And far too many ordinary people have no lawyer at all. The access-to-justice problem is vast, and our existing efforts have proven woefully inadequate. We mustn’t make the best the enemy of the good. It’s time to explore the future responsibly instead of freezing our broken status quo in place.

Conclusion

As Niels Bohr deadpanned: “It is very difficult to predict, especially the future.”60 The legal-services market, for so long a cartel, is ripe for disruption. Paraprofessionals and streamlined small-claims procedures offer alternatives for smaller matters and people of modest means. Technology, especially AI, can expand legal assistance and drive costs way down. That promises to democratize justice, helping those who have long taken their lumps and done without help.

The broader lesson here is one of institutional design. Courts have long been designed around a set of adversarial assumptions, with reticulated procedures and technical jargon that only lawyers can navigate. The Federal Rules of Civil Procedure are trans-substantive, forcing disputes huge and tiny into the same Procrustean bed. Much of the problem is of courts’ own making. By simplifying procedures and allowing experimentation, courts can create space for tech-fueled access to justice.

Stephanos Bibas is a judge on the U.S. Court of Appeals for the Third Circuit and a Senior Fellow at the University of Pennsylvania Law School. He thanks Nathan Pinnell, Arjun Prakash, and Isabella Soparkar for their excellent help.