By Jeffrey T. Donner, Esq.

June 23, 2026

Generative AI has not made good lawyers obsolete.

It has not replaced judgment. It has not replaced experience. It has not replaced credibility, negotiation, courtroom skill, client counseling, or the professional responsibility that comes with signing your name to a filing.

A lawyer still has to know what matters and what does not. A lawyer still has to know what must be verified, what should be abandoned, what argument is dangerous, what fact is unsupported, and what a judge is likely to care about.

But AI has already changed something important.

It has destroyed a lot of traditional billable-hour inventory.

That is the part the legal profession needs to discuss more honestly.

For decades, much of the law-firm business model rested on the fact that legal work took time. Research took time. Drafting took time. Chronologies took time. Summaries took time. Deposition outlines took time. A young associate might spend eight, ten, or twenty hours preparing a research memorandum or document summary. A senior lawyer might then spend half an hour or an hour reviewing it, revising it, and deciding what to do.

The client paid for both parts: the associate’s production time and the senior lawyer’s judgment.

Generative AI changes the production-time part dramatically.

A capable lawyer can now use AI to generate a first-pass chronology, deposition outline, argument structure, case summary, or research framework in minutes. Sometimes seconds. Sometimes the first draft is bad. Sometimes it is incomplete. Sometimes it is dangerously confident and wrong.

But sometimes it is very good.

That does not mean the lawyer’s work is finished. The lawyer still has to check the cases. The lawyer still has to compare the output against the actual record. The lawyer still has to identify missing facts, remove weak arguments, correct overstatements, and decide what should actually be used. The lawyer still takes responsibility.

But the first architecture of the work product may arrive almost instantly.

That is not a small efficiency gain. In some tasks, it is a revolution.

And it creates a direct collision with hourly billing.

If a lawyer bills by the hour and uses AI to complete in thirty minutes what used to take eight hours, what should the lawyer bill?

The answer should be simple: the lawyer should bill the actual time reasonably spent. That includes the time spent prompting, reviewing, verifying, revising, checking citations, applying judgment, and taking responsibility for the final product.

If that is thirty minutes of drafting and another hour of review, then that is the time.

What the lawyer should not do is secretly bill the eight hours the task might have taken in the pre-AI world.

That is not hourly billing.

That is undisclosed value billing.

There is nothing inherently wrong with value billing. Flat fees, contingent fees, success fees, premium rates, and other alternative fee arrangements can be legitimate. A client may agree to pay a fixed amount for a defined result, a defined deliverable, or a defined phase of litigation. If the lawyer then uses skill, experience, and technology to perform the work efficiently, the lawyer may benefit from that efficiency.

But that has to be the deal.

If the agreement is hourly, the bill should be hourly. If the agreement is value-based, the client should know that up front.

The Florida Bar has already recognized this issue. In Florida Bar Ethics Opinion 24-1, the Bar states that lawyers may use generative AI, but must protect client confidentiality, remain responsible for work product, verify accuracy, avoid improper billing practices, and comply with the rules governing legal fees. The opinion specifically warns that increased efficiency from generative AI must not result in falsely inflated claims of time. It also recognizes the alternative: lawyers may consider flat fees or contingent-fee arrangements so that the benefits of increased efficiency accrue to both lawyer and client.

The ABA has said much the same thing. In ABA Formal Opinion 512, the ABA explains that lawyers using generative AI remain responsible for competence, confidentiality, communication, supervision, candor, and reasonable fees. On hourly billing, the opinion is direct: lawyers must bill for actual time. If AI helps draft a pleading and the lawyer spends fifteen minutes entering information, the lawyer may bill for those fifteen minutes and for the actual time spent reviewing the draft. The lawyer may not bill for the longer amount of time the task might have taken without AI.

That is the clean line.

AI does not require lawyers to work for free. It does require honesty about what fee arrangement is being used.

Hourly billing has always had a tension built into it. Clients want efficiency. Hourly billing rewards time. A lawyer who solves a problem quickly may create more value than a lawyer who takes all day, but the traditional hourly model pays the second lawyer more.

AI makes that tension much harder to ignore.

This does not mean clients should use AI instead of lawyers. That would often be a serious mistake.

AI can invent authority. It can miss controlling law. It can misunderstand a contract. It can summarize a document correctly but draw the wrong conclusion. It can provide generic advice that sounds plausible but fails in the actual procedural posture of a case. It can give a client confidence without giving the client judgment.

The best use of AI is not client self-lawyering. It is a competent lawyer using AI as a force multiplier.

That is how I use it in my own practice. I use modern tools to work faster, organize facts more efficiently, and get to the legal issues sooner. But I still verify the work. I still review the record. I still check the cases. I still decide what arguments should be made and which ones should be abandoned. I still take responsibility for what is filed or sent.

And when I bill hourly, I bill the real attorney time spent doing that work.

That should matter to clients.

A client should want a lawyer who uses modern tools well. But the client should also want a lawyer who is honest about the difference between time and value. If technology allows a lawyer to perform a task more efficiently, an hourly client should receive the benefit of that efficiency. If the lawyer wants to charge for value instead, the lawyer should say so in the fee agreement.

The lawyer is not being paid merely to type words.

The lawyer is being paid to know which words matter and whether they should be used at all.

Still, the profession should not pretend nothing has changed.

If AI reduces the time needed to prepare a first draft, a chronology, a deposition outline, or a research framework by 80 or 90 percent, then a lawyer billing hourly cannot ethically pretend the time was still spent. The fact that the final product has value does not turn imaginary time into billable time.

If the lawyer wants to charge for value, the lawyer should say so.

That may be where much of the profession eventually goes. Some lawyers will continue to bill by the hour and pass efficiency savings to clients. Some will move toward flat fees for defined work. Some will use hybrid arrangements: hourly billing for open-ended litigation, flat fees for discrete motions, fixed-fee phases for discovery or hearings, or contingent components where permitted.

Those models may make more sense in an AI-assisted world because they align the lawyer’s incentive with efficiency.

But there is a difference between changing the fee agreement and quietly changing the meaning of an hour.

The free market does not owe lawyers the same income they made before a productivity revolution. If technology allows one experienced lawyer to do work that once required a partner, two associates, and a paralegal, clients will eventually ask why the bill still looks the same.

That is a fair question.

The answer cannot be that lawyers are entitled to preserve the old economics because the old economics were comfortable. Nor can the answer be that AI-assisted legal work should become nearly free. The better answer is that clients should pay for legal judgment, responsibility, risk, skill, and results, while lawyers should be candid about whether they are charging for time or value.

Lawyers who adapt well will not be the ones who pretend AI is just a fancy spell-checker. They also will not be the ones who treat AI as a substitute for legal judgment.

The lawyers who adapt well will be the ones who can say: I use modern tools to work more efficiently. I verify the work. I exercise judgment. I remain responsible. And my fee arrangement tells you honestly what you are paying for.

If the representation is hourly, the client pays for real time.

If the representation is flat-fee or value-based, the client agrees to that structure in advance.

AI has not killed lawyers.

But it has exposed a lot of billable friction.

The profession should be honest about that before clients figure it out for themselves.