Hot Take: AI Won't Shrink Your Litigation Bill. It'll Make the Fight Bigger.

Every legal AI pitch ends with the same slide: lower costs, faster turnaround, happier GC. For your NDAs, maybe. For your next big lawsuit? Grab a coffee.

On October 12, Artificial Lawyer founder Richard Tromans proposed what he calls the "Legal AI Law of Adversarial Incentives": the more the parties want to fight, the more AI efficiency gains create additional work instead of saving it. Low-conflict work gets cheaper. High-conflict work gets bigger. Meanwhile, the past week's launches handed both sides of every dispute a lot more ammunition.

Exhibit A: what shipped last week

• Harvey and LexisNexis launched an agentic motion-to-dismiss workflow that identifies up to eight candidate grounds, rates their strength, and drafts a citation-backed motion. Harvey says an MTD typically takes 40 to 50+ hours of associate work. (Artificial Lawyer, October 9)
• Cooley and Google are building a Gemini Enterprise agent that recommends redactions in court filings for attorney review. (Artificial Lawyer and Legal IT Insider, October 9)
• OpenAI Decisions, in public beta, classifies content at volume for $0.10 per million input tokens. Artificial Lawyer flagged first-pass eDiscovery and due diligence review as the obvious legal use. (Artificial Lawyer, October 8)

All useful. All available to the other side too.

Seven hard truths for your litigation budget

1. A cheaper motion is a more likely motion

When a motion to dismiss took 50 hours, counsel thought twice. When an agent drafts the first version, "let's file and see" becomes a lot easier to say.

2. Your opponent read the same press releases

AI doesn't give you an edge when both sides have it. It raises the baseline. The fight doesn't get shorter. It gets more thorough.

3. Bulk review for pennies makes "review everything" a real strategy

Once first-pass classification costs almost nothing, expect discovery requests to get broader and the "unduly burdensome" objection to get harder to win.

4. Your contract work and your litigation work are heading in opposite directions

Tromans' own examples make the point: a startup's contract review gets the full efficiency gain, while a large corporate dispute absorbs it. If your budget treats legal AI savings as one number, it's already wrong.

5. The budget won't shrink. It will move.

Hours saved on drafting turn into hours on strategy, experts, and more motions. Ask your e-billing data where the hours went, not just whether they dropped.

6. The biggest AI savings are in fights you never start

Standard terms, clear contracts, and quick early assessment do more for your litigation spend than any drafting agent. Low conflict is where AI pays off, so create more of it.

7. "What will you do with the 40 hours?" is now a fair question

If your firm's AI really cuts a motion from 50 hours to 10, you're allowed to ask whether you'll see that on the invoice, or whether it turns into 30 more hours of "strategic refinement."

What to actually do about it

• Split your AI savings forecast: transactional and advisory work in one bucket, disputes in another. Expect different results.
• Invest in early case assessment. eDiscovery and ECA tools help you decide early which fights are worth having.
• Set expectations with outside counsel: ask how AI changes staffing, motion practice, and pricing on each new matter.
• Prepare for bigger discovery asks. Clean data and solid legal holds make the "review everything" era less painful. The litigation and investigations category is a good place to start.

AI makes legal work cheaper per unit. Litigation has never been short on units.

This article is general information, not legal advice.