The standard narrative about AI in legal is that it threatens lawyers. The first-year associates lose drafting work. The document review industry collapses. Partners lose hours. The pie shrinks.
I want to put forward a different argument from a conversation I had with Charles Y. Davis, managing partner of Brown, Hay & Stephens, a firm nearly 200 years old and adept at adapting to changing times.
“The cost of litigation is a gatekeeper. Ask someone to put up $25,000 or $100,000 on a case they might not win, and a lot of folks just walk away. This doesn’t mean the problems aren’t real, it means the price to get in is too high for them.
Charles Y. Davis, Managing Partner
If AI meaningfully brings the cost down, it stands to reason you’ll see more cases filed. Maybe quite a few more over time, depending on how it plays out. I don’t think this means fewer lawyers, but instead likely means lawyers working at a different level, handling a large volume of cases that used to be out of reach.”
Charles is right. The data supports him better than the prevailing narrative does.
How big is the gap
Let’s start by looking at what’s actually keeping cases out of the courthouse.
The Legal Services Corporation’s 2022 Justice Gap report found that 92% of substantial civil legal problems faced by low-income Americans receive no or inadequate legal help. The single most-cited reason is cost. The civil legal aid funding shortfall is about $1.4 billion a year.
It’s not just low-income. The National Center for State Courts estimates that taking a contract dispute through trial can cost about $91,000. Even mid-market businesses sit out claims they would clearly win because the math does not work. Based on Clio’s 2025 data, the average lawyer billing rate of $349 an hour puts a 40-hour pre-trial workup at nearly $14,000 before a case is ever filed.
Richard Susskind has been making a version of this point for years. He calls it the “latent legal market.” In a 2023 interview with Legal Futures, he argued that there is a vast latent legal market of people with legal problems who receive no legal help because lawyers are too expensive. The demand for legal help is not what we see in courthouses. What we see in courthouses is the demand that can afford the gatekeeper. The actual demand, including everyone priced out, is several times larger.
That is the population AI is about to make economically viable.
What history says happens when the barrier drops
Three precedents matter here.
E-discovery
This is the closest analog because it happened within the same profession under the same “AI replaces lawyers” narrative. After Zubulake (2003 to 2005) and the rise of predictive coding around 2012, document review fell dramatically. Manual review could cost several dollars per document, while predictive coding reduced the expense to a fraction of traditional review costs. Everyone said document review jobs were finished.
What actually happened was the opposite.
According to Fortune Business Insights, the e-discovery market reached $18.73 billion in 2025 and is projected to hit $46.06 billion by 2034. The managed review industry did not exist before 2003. Contract attorney staffing firms did not exist before 2003. The e-discovery counsel subspecialty inside the AmLaw 200 did not exist before 2003. Lower unit cost expanded the universe of reviewable data and created an entirely new supervisory layer.
Class actions
In 1966, the federal rules were amended to allow opt-out class actions for small-dollar claims. Before that, a $50 consumer harm spread across a million customers was uneconomic to pursue. The amendment changed that. Much of the modern plaintiffs’ bar traces its roots to this shift, including firms such as Hausfeld, Lieff Cabraser, and Robbins Geller. Securities class actions alone settled $3.7 billion in 2024.
One procedural rule change collapsed the per-plaintiff cost barrier. Out came an entire new category of litigation and an entire new specialty bar to supervise it.
ADA Title III since 2013
This one is happening right now, and it’s closer to the AI scenario than the other two. Federal ADA Title III lawsuits grew from 2,722 in 2013 to 8,667 in 2025. Industry estimates suggest 35,000 to 50,000 ADA-related website accessibility demand letters are sent annually, far exceeding the number of filed cases. Serial-plaintiff firms supervising this volume have scaled up, not been disintermediated.
The pattern is consistent across all three. Lower the cost barrier. Volume explodes. A new supervisory layer of lawyers emerges to manage the volume. Total legal demand goes up. Total lawyer demand goes up.
The Jevons paradox applies to law
Economists have a name for this. The Jevons paradox describes what happens when technology makes a resource cheaper to use. Total consumption rises faster than the per-unit cost falls. Coal was the original example. Each engine become more efficient, and total coal use went up.
Legal services have shown the same dynamic every time we have measured it. E-discovery, class actions, ADA filings. The cheaper the marginal complaint, the more complaints get filed.
Susskind put it plainly in the same 2023 interview, arguing that AI threatens the artificial scarcity that has long defined the legal market. When the scarcity goes, the market that was hidden behind it shows up.
Addressing the obvious objections
A few rebuttals worth handling head-on.
Federal civil filings dropped 14% in 2024.
True. But the drop was almost entirely the wind-down of one mass tort docket (the 3M earplug litigation). Strip the multidistrict litigation distortion and federal civil litigation has been stable. State courts, where 98% of US civil litigation occurs, handled around 70 million cases last year, and civil case loads were up 5%. The cost barrier is binding, not loosening.
Online dispute resolution will absorb the volume.
Partially right. The UK’s Online Civil Money Claims system has processed more than 470,000 claims, while Utah’s small-claims platform has demonstrated similar success with online dispute resolution. But ODR works for sub-$5,000 disputes. The litigation Charles Y. Davis is talking about, the kind that drives mass torts, securities cases, and complex commercial work, is exactly the supervision-heavy work that AI volume creates more of, not less.
Tort reform will throttle it.
Reform always comes. The American Invents Act capped patent troll cases. UK whiplash caps cut volume in half. Texas medical malpractice reform reduced med-mal filings by 60%. But reform takes 5 to 15 years, and the supervisory infrastructure built in the meantime persists. Plus, the Texas data shows reform redistributes lawyer work rather than eliminating it. Plaintiffs’ firms moved into commercial torts and consumer cases. Total state civil filings barely moved.
Lawyers will be replaced at the supervisory layer too.
Maybe eventually. Right now, every US state’s unauthorized practice rules require a licensed attorney of record. Mata v. Avianca and the wave of AI-hallucination sanctions are making courts more reliant on supervising counsel, not less. The regulatory moat is still there.
What this means for firm strategy
If Charles is right, the firms that win the next decade are the ones that figured out how to supervise litigation at scale. That is a different operating model than the one most firms are built on.
It looks more like:
- Heavy use of AI for intake, screening, drafting, and document handling
- A smaller number of senior lawyers exercising judgment, with each senior lawyer’s effective capacity multiplied 10x or more
- Pricing is built on volume and outcome, not hours
- Operations infrastructure that looks closer to a managed-services business than a traditional firm
The plaintiffs’ bar has been building toward this for a decade. Mass tort firms now manage case inventories that have doubled since 2018. The top 50 firms employ about 12,000 lawyers and 40,000 staff. That is a supervisory model already running at an industrial scale.
Defense firms, regulatory practices, and in-house departments are next. If filings grow the way the historical analogs suggest they will, the firms that built supervisory capacity will absorb the work. The firms that did not will lose share to those that did, or to AI-native challengers operating on a per-output pricing model.
The compressed version
The AI-replaces-lawyers narrative is looking at the wrong end of the market. AI is going to lower the cost barrier that has been keeping most legal grievances out of court for as long as we have been measuring. When that barrier comes down, history says the volume of lawsuits rises faster than the per-unit cost falls. That volume needs to be supervised.
In sum, the next decade is not going to be defined by fewer lawyers. It will be defined by lawyers working at a different level, on a much larger pile of work, in firms organized very differently than they are today.
Charles called this and might call it again in nine years. As he put it himself when he made the case, “I’m wrong about a lot of things.” Maybe. But the data is lining up with him.
This publication contains general information only and Sikich is not, by means of this publication, rendering accounting, business, financial, investment, legal, tax, or any other professional advice or services. This publication is not a substitute for such professional advice or services, nor should you use it as a basis for any decision, action or omission that may affect you or your business. Before making any decision, taking any action or omitting an action that may affect you or your business, you should consult a qualified professional advisor. In addition, this publication may contain certain content generated by an artificial intelligence (AI) language model. You acknowledge that Sikich shall not be responsible for any loss sustained by you or any person who relies on this publication.
