This is an opinion piece. It reflects the views of Andrew M. Crawford about where the legal market is heading. The dollar figures in the examples are illustrations, not quotes or results.

Everyone says AI will kill litigation. They have it backwards. AI is about to make litigation explode.

For a hundred years, the hourly rate was the dam

American litigation has always been limited by one thing more than any other: the hourly rate. Not the merits. Not the law. The cost of paying someone to fight.

  • A breach of contract worth $40,000? The lawyer estimates $25,000 to fight it. You settle for $12,000 and call it a win.
  • A customer stiffs you for $8,000? Not worth the demand letter. You eat it.
  • A former employee walks off with your client list? Your lawyer says, “We can send something, but the motion practice alone could run $30,000.” You let it go.

That is how most American disputes end. Not in court. In a spreadsheet, where someone decides the fight costs more than the loss. The hourly rate has quietly filtered out an enormous number of legitimate claims, and an enormous number of bad ones, for as long as any of us has been alive.

Now change one number

A demand letter that used to take hours now takes minutes. A first draft of a motion that used to consume days of associate time now takes a fraction of that. A long discovery response that took a paralegal two weeks to assemble can be drafted overnight.

The judgment still matters, and a lawyer still has to check every word, because courts have now dealt with AI-invented citations in well over a thousand cases, and imposed sanctions in hundreds of them. But the cost of producing legal paper has collapsed, and it will keep falling.

So what happens to the spreadsheet? Every “not worth it” flips to “why not.”

Jevons paradox comes to the courthouse

Economists have a name for this. In 1865, William Stanley Jevons noticed something strange about the steam engine. As engines became more efficient, everyone expected Britain to burn less coal. Britain burned more. Cheaper power meant more factories, more railways, and uses nobody had planned for.

It happens again and again. Bandwidth got cheap, and we did not send fewer emails. We built streaming video. Computing got cheap, and we did not use fewer computers. We put one in every pocket.

Legal drafting is getting cheap. People will not fight less. They will fight more.

Cooperative work shrinks. When both sides want the same standard nondisclosure agreement, AI hands it over in a minute and the deal closes. A large part of routine transactional drafting is going to get faster, cheaper, and in some cases disappear.

Adversarial work grows. The moment your opponent has AI, every letter they send demands a response from you. Every motion breeds a response, and a reply, and a cross-motion. There is no natural ceiling on an arms race when the marginal cost of the next round falls toward zero for whoever fires first.

Then it hits a wall: judges

AI can multiply the filings. It cannot multiply the judges. In the federal district courts, civil cases that go to trial commonly take years to get there, and many state courts are just as strained. The early warning signs are already on the docket: a public database of court decisions involving AI-fabricated citations has passed 2,000 entries worldwide, most of them in the United States and a large share from people representing themselves. Judges have responded with standing orders, including one from a federal judge in the Northern District of Texas requiring lawyers to certify how generative AI was used in their filings.

Texas is adding capacity, including a specialized Business Court that began hearing cases in 2024. But picture the arithmetic: many times the motions, roughly the same number of judges, clerks drowning, and dockets stretching further out. Something breaks. I do not know what. But it will not be boring.

What this means for businesses and regular folks: the fight gets asymmetric

Here is the part that matters most for the people Crawford Law serves. The cost of starting a fight is collapsing much faster than the cost of defending one.

  • A former employee can generate a polished demand letter, an agency charge, and a draft complaint in an evening.
  • An unhappy customer can send a ten-page letter citing statutes, threaten a lawsuit, and post the whole thing online.
  • A competitor or a serial filer can send the same claim to fifty businesses at once and see who pays.
  • A self-represented party can file motion after motion that each look professional and each require a response.

On their side, it is a few prompts. On yours, it is a deadline, a business to run, employees and customers to protect, an insurer to notify, and real risk if you guess wrong. Texas deadlines do not care how a filing was written. In county and district court, an answer is generally due by 10 a.m. on the Monday after 20 days from service, and in justice court debt cases by the end of the 14th day. Miss it, and a default judgment can follow. A cheap filing can be very expensive to ignore.

Why you can’t just answer prompts with prompts

You can write a few prompts too. Sometimes that will be enough. But you have your business to protect, and a machine-written response creates its own risks:

  • It can be wrong in ways you will not catch. Invented cases, misstated deadlines, and the wrong legal standard read just as confidently as the right ones.
  • It can say too much. An eager response can admit facts, waive defenses, or hand the other side the theory of its case.
  • It may not be confidential. What you type into a consumer AI tool may not be protected by the attorney-client privilege.
  • It does not know your obligations. Preserving evidence, notifying an insurer, and following procedural rules are easy to miss and expensive to get wrong.
  • It is not accountable. A lawyer signs, owes you duties, and answers to the court and the bar for what is filed.

What a rapid, economical response looks like

  1. Do not ignore it. Note the date you received it and calendar every deadline.
  2. Preserve the evidence. Stop routine deletion of emails, texts, and files that relate to the dispute.
  3. Check your insurance. Employment practices, general liability, and other policies often require prompt notice and may provide a defense.
  4. Get a fast, fixed-price read on the merits. Is the claim real, how big is the exposure, and what does each path cost?
  5. Right-size the response. Sometimes no response, sometimes a short firm letter, sometimes a negotiated resolution. When a lawsuit is baseless, Texas offers early tools: a motion to dismiss under Rule 91a must be filed within 60 days and may lead to an award of fees; claims over speech can face a Texas Citizens Participation Act motion; and courts can sanction groundless pleadings and restrict vexatious litigants.
  6. Settle when the math says settle. Speed and cost discipline make settlement a choice, not a surrender.
  7. Get ahead of the next one. Contracts with clear dispute resolution, venue, and fee-shifting terms; consistent documentation of discipline and investigations; and policies that are followed. The best defense against an asymmetric fight is a record that makes the claim not worth bringing.

Where Crawford Law fits

AI should make good legal work faster and cheaper for clients too, and Crawford Law uses it the way the Texas ethics guidance expects: as a tool, with a lawyer responsible for every word. That efficiency is what makes rapid, economical responses possible. A demand letter, a reply to one, a response to an agency charge, an early motion, an arbitration, or a defamation dispute can be triaged quickly and quoted up front as a fixed block, so you know the cost of the fight before you are in it. When a case needs a dedicated trial team, Andrew will tell you, and help you find the right one.

The other side can write a few prompts. You have a business to protect. Request a free call, or see the business practice and workplace practice.

A note for young litigators

If you are early in your career, the paperwork was never going to be your career. The machine has that now. The growth is in the fight. Learn to take a deposition. Learn to find the hole in a 60-page brief a machine wrote in ten minutes. Learn to stand in front of a judge with hundreds of cases on her docket and make yours the one she remembers.

The machine drafts. Someone still has to win. Make sure it is you.

Sources

  1. W. Stanley Jevons, The Coal Question (1865)
  2. Administrative Office of the U.S. Courts, Federal Court Management Statistics and Table T-3 (median time from filing to trial, civil cases)
  3. Damien Charlotin, AI Hallucination Cases database, as summarized in 2026 trackers (more than 2,000 decisions worldwide)
  4. Mata v. Avianca, Inc., 678 F. Supp. 3d 443 (S.D.N.Y. 2023) (sanctions for fabricated citations)
  5. Judge Brantley Starr, N.D. Tex., Mandatory Certification Regarding Generative Artificial Intelligence (2023)
  6. Professional Ethics Committee for the State Bar of Texas, Opinion 705 (2025) (lawyers’ use of generative AI)
  7. Texas Rule of Civil Procedure 91a (dismissal of baseless causes of action)
  8. Texas Citizens Participation Act, Texas Civil Practice & Remedies Code ch. 27
  9. Texas Civil Practice & Remedies Code ch. 10 (sanctions for frivolous pleadings) and ch. 11 (vexatious litigants)
  10. Texas Rules of Civil Procedure 99 and 502.5 (answer deadlines)
  11. Texas Government Code ch. 25A (Texas Business Court, operating since Sept. 1, 2024)