By Jeffrey T. Donner, Esq.

September 9, 2026

I have been reading about the coming death of the billable hour for almost as long as I have been practicing law.

When I began practicing in 1999, Law.com was among the relatively new legal publications I regularly read online. Even then, articles appeared predicting that corporate clients were finally going to rebel against hourly billing, that alternative fee arrangements would replace it, and that law firms would have to learn to sell value rather than time. The arguments were familiar: hourly billing rewards inefficiency, makes legal expenses unpredictable, and creates an inherent tension between a client who would like a task completed quickly and a lawyer or law firm that earns more revenue when the task takes longer.

More than a quarter-century later, the billable hour remains very much alive. That persistence should make us cautious about periodically announcing its imminent death. Hourly billing has significant shortcomings, but it also addresses a genuine economic problem: in many kinds of legal work, and particularly contested litigation, nobody can know at the beginning how much professional labor will ultimately be required.

Artificial intelligence does not eliminate that problem. What it does is change an important part of the equation by dramatically reducing the human labor required for certain categories of legal work. That development is exposing contradictions in law-firm economics that have existed for decades, while simultaneously creating an opportunity for smaller firms to provide sophisticated legal services more efficiently.

The difficulty is that two different groups are caught in the transition. Clients understandably want the economic benefit of technological efficiency, particularly when they must finance litigation out of their own pockets. At the same time, young lawyers at traditional firms may still be required to generate 1,900, 2,000, or more billable hours each year from a shrinking inventory of work that legitimately takes that long. Neither side created the problem, and neither side can solve it by pretending that professional time no longer matters.

The Billable Hour Has Always Had an Efficiency Problem

The traditional criticism of hourly billing is mathematically simple. If two equally capable lawyers reach the same correct answer and one requires twelve hours while another requires six, an hourly billing system generates twice as much revenue from the less efficient process.

Experience therefore can create a strange economic result. A lawyer who has practiced for decades may immediately recognize that three possible arguments are dead ends and that only one deserves serious research. A younger lawyer encountering the issue for the first time may spend ten hours investigating all four theories before reaching the same conclusion. The senior lawyer has delivered the answer more efficiently because of accumulated knowledge, but a pure hourly system may reward the longer route.

This is not necessarily evidence of wrongdoing or inefficiency by the younger lawyer. Lawyers have to learn. Difficult questions genuinely take different lawyers different amounts of time, and legal problems are not standardized industrial products. The point is simply that hours are an imperfect proxy for professional value.

Large firms historically addressed this problem through leverage. Senior lawyers identified issues and directed strategy, while junior associates performed much of the labor-intensive research, factual organization, document review, and initial drafting. The senior lawyer then reviewed the work, corrected it, applied experience and judgment, and communicated with the client. The client paid for both layers because both layers involved real professional labor.

For much of modern legal practice, that model made economic sense. The junior work actually had to be performed by a human being, and it often consumed substantial amounts of time. Artificial intelligence is changing that first layer more rapidly than anything I have seen during my career.

I Remember When Legal Research Actually Took That Long

I entered practice in 1999 and spent significant portions of my earlier career at Holland & Knight, Gunster, and Podhurst Orseck. I therefore experienced the traditional large-firm research model firsthand.

When a senior lawyer encountered a difficult legal issue, an associate might be assigned to research it and prepare an interoffice memorandum. That could require many hours identifying authorities, reading appellate opinions, following citations, checking the subsequent history of cases, comparing conflicting decisions, organizing the law, and writing a coherent explanation for the lawyer supervising the matter. A complicated assignment could consume several working days.

When I began practicing, some of that research still involved actual books. Westlaw and other electronic research systems were already indispensable, but the physical law library remained part of law-firm life. I remember using bound reporters and treatises as well as spending many hours electronically searching cases, following citations, copying pertinent language, and incorporating the research into briefs and memoranda.

There were periods when I billed ten-hour days doing this work. The time was billed to clients because I actually worked those hours. It was not make-work. Someone had to locate and understand the law before the firm could responsibly give the client an answer.

The senior lawyer then had additional work to perform. A partner could not responsibly accept an associate memorandum without understanding the underlying issue. The partner had to read the work, review significant authorities, identify weaknesses or missing questions, synthesize the research with the particular facts, and determine what advice should be given or what argument should be made.

The client therefore paid for two legitimate layers of professional labor. Depending upon the issue, fifteen or twenty associate hours might precede several additional hours of senior-lawyer review. That was not inherently excessive; it reflected how the work was actually performed with the available technology.

Artificial Intelligence Compresses the First Layer of Professional Labor

Modern generative AI can now perform remarkable portions of that first-stage work almost instantaneously. A properly framed research question can produce a sophisticated preliminary memorandum identifying potentially relevant cases, competing arguments, factual distinctions, procedural questions, and issues requiring further investigation. AI can search lengthy contracts and insurance policies, organize chronologies, compare documents, summarize complicated records, and produce an initial structure for analysis far faster than a human lawyer working manually.

That does not mean a five-minute AI memorandum is a finished legal product. A responsible lawyer still has to verify the authorities, read the material cases, confirm that the jurisdiction is correct, understand the procedural posture, identify adverse precedent, determine whether the cases remain good law, examine the precise statutory or contractual language, compare the law with the actual factual record, and decide whether the analysis is sound.

Nevertheless, the time savings are real. An assignment that could reasonably have consumed fifteen hours of junior-associate time in 1999 may sometimes require only several hours of actual lawyer time today. In other matters AI may save five hours rather than twelve, and in some assignments it may save very little because the difficult part of the work is not locating information but deciding what the information means.

This is why I think it is misleading to say that AI simply “does the lawyer’s work.” It changes the allocation of the work. Less professional time may be required to locate, initially organize, type, and summarize information, while the remaining time becomes more heavily concentrated on verification, analysis, strategy, and judgment.

That can improve quality as well as efficiency. A lawyer who spends fewer hours mechanically searching for information can spend more of the available professional time asking whether the apparent answer is actually correct, what the opposing party will argue, what evidence remains missing, and what should be done next.

The Senior Lawyer Still Has to Read, Verify, Understand, and Decide

Even when AI produces a first draft that is exceptionally good, the lawyer with professional responsibility for the matter still has to sit down and read it.

That obligation is not ceremonial. Apparently straightforward legal propositions often contain important qualifications. Jurisdiction matters. Procedural posture matters. Burdens of proof matter. Precise policy and contractual language matter. An appellate court may state a broad proposition while deciding the particular appeal on substantially narrower grounds.

Modern AI systems are increasingly capable of recognizing those distinctions themselves. I have seen them identify jurisdictional limitations, distinguish procedural postures, compare policy language, and locate factual distinctions with considerable sophistication. The appropriate response is not to pretend that the technology is less capable than it is. The appropriate response is to recognize that the lawyer remains professionally responsible for determining whether the analysis is correct.

That means returning to primary authorities and traditional research tools, reading the significant opinions, examining the actual documents, testing the proposed reasoning, and determining whether something material has been overlooked. It also means counseling the client, which is fundamentally different from generating legal information. Someone has to decide how much the uncertainty matters, what additional work is economically justified, whether settlement makes sense, and what risks the client should accept.

Those hours remain real billable work. If artificial intelligence saves five hours of preliminary research but an experienced lawyer still needs six hours to read, verify, understand, apply, revise, and advise, it is incorrect to conclude that the six hours were unnecessary because “AI did the research.” AI eliminated some labor. It did not eliminate professional responsibility.

The ABA and The Florida Bar Are Now Confronting the Same Problem

The profession’s governing and professional institutions have begun addressing this collision between technological efficiency and traditional law-firm economics.

The American Bar Association’s Law Practice Division recently published an article titled “Rebuilding Law Firm Economics Starts with What Clients Buy.” Its central point is that AI is exposing structural defects in the traditional law-firm model rather than creating them. The article reports declining billable hours at many firms while billable-hour targets continue to rise, and it argues that the traditional law-firm pyramid is under pressure because AI is compressing precisely the junior-level execution work on which that pyramid historically depended.

The ethical rule is clearer than the business solution. ABA Formal Opinion 512 states that a lawyer billing hourly must bill actual time spent. If generative AI allows the lawyer to complete a task substantially faster, the lawyer cannot bill the client for the hours that would hypothetically have been required without AI. The lawyer may bill for the actual time spent using, reviewing, correcting, and applying the AI-assisted work, but efficiency does not create fictional time.

The Florida Bar reached substantially the same conclusion in Ethics Opinion 24-1. A lawyer may not use generative AI to duplicate charges or falsely inflate billable time. The Florida opinion also recognizes, however, that flat fees, contingent arrangements, and other structures may provide ways for lawyers and clients to share the economic benefit of increased efficiency.

That is the unresolved problem. The ethical rule says that technology-created efficiency must benefit the hourly client through fewer hours. The traditional law-firm business model may simultaneously expect the lawyer to replace every hour technology eliminates with another billable hour somewhere else.

Eventually those two systems have to be reconciled.

The Young-Associate Problem Is Not Hypothetical

Young lawyers entering large firms today occupy a particularly difficult position because they are being trained under economic assumptions that predate the technology they use every day.

Many 24-, 25-, and 26-year-old associates used generative AI during the latter part of law school and entered practice already comfortable using it to research, organize, draft, summarize, and critique written work. To them, the technology is not an unusual innovation requiring a special committee meeting. It is part of how intellectual work is done.

At the same time, many firms continue to measure associates against annual targets of approximately 1,900, 2,000, or more billable hours. Those quotas were difficult even when research memoranda regularly consumed an entire day or several days. Not every hour spent at work is billable, and lawyers also spend time on training, firm administration, professional development, marketing, mentoring, and other activities that do not appear on client invoices.

AI is now removing part of the inventory from which those quotas historically were filled.

I personally know practicing lawyers who have told me that they sometimes record substantially more billable time for an assignment than the AI-assisted work actually consumed. I am intentionally not identifying any lawyer or firm, and I am not suggesting that this is universal practice. I am describing conversations I have personally had with lawyers currently practicing in firms.

The incentive is not difficult to understand. Assume that a young associate has a 2,000-hour expectation and receives a research assignment that would have consumed twelve hours twenty-five years ago. The associate uses AI intelligently, verifies the cases, reviews the authorities, revises the memorandum, and finishes the assignment properly in three hours. The client has benefited from nine hours of technological efficiency, but the associate has also lost nine hours toward a quota that remains unchanged.

Repeat that process throughout a year and the tension becomes obvious. The temptation to record what the assignment “would have taken” using older methods may be understandable, but it is not honest hourly billing. If the lawyer worked three hours, the lawyer worked three hours.

This resembles the older problem sometimes called unit billing: recording a predetermined quantity of time for a task rather than the time actually spent performing it. Generative AI did not invent that ethical problem. It simply creates a new and potentially much larger opportunity for it.

Law Firms Cannot Demand Efficiency and Then Punish Associates for Achieving It

The solution should not be to blame young lawyers for learning to use technology effectively.

If a twenty-six-year-old associate can perform in three hours an assignment that reasonably took me fifteen hours when I was twenty-six, and if the young lawyer independently verifies the research and produces an equally reliable work product, that lawyer has performed efficiently. The client should receive the benefit of the saved hourly time.

The firm, however, cannot fairly demand technological efficiency while simultaneously insisting that the associate somehow reproduce the missing twelve billable hours. If the underlying quantity of legitimate human labor changes, eventually the economic measurement of lawyer productivity has to change with it.

That may mean lower hourly targets. It may mean smaller associate classes. It may mean greater use of flat fees, portfolio pricing, hybrid arrangements, success fees, or other structures in which efficiency increases profitability rather than reducing it. It may also mean changing the way young lawyers are evaluated so that judgment, quality, client service, technological proficiency, and actual contribution matter more relative to raw hours.

There is also an apprenticeship problem. The fifteen-hour memorandum of the old system did more than generate revenue. It trained the associate. Reading twenty cases, tracing the development of a doctrine, and struggling to organize the analysis taught young lawyers how law actually works.

Clients should not be forced to finance deliberate inefficiency merely because law firms need to train associates. Firms will have to find new ways to teach the underlying skills of verification, reasoning, judgment, drafting, and advocacy in an environment where technology eliminates much of the repetitive work through which previous generations learned them.

Not All Legal Work Can Be Compressed

It is equally important not to exaggerate what AI does to the overall inventory of legal work.

An eight-hour deposition remains an eight-hour deposition. A full day in court remains a full day in court. A lawyer trying a case for four or five weeks may legitimately bill very long days because the lawyer is physically present in the courtroom, examining witnesses, reviewing exhibits, researching unexpected issues, meeting with the client, preparing the next day’s examinations, and responding to an adversary in real time.

I experienced substantial stretches of that kind of work early in my career. During an extended trial there can be no shortage of legitimate hours. The lawyer may be in court all day and working much of the night preparing for the next day.

Hearings, mediations, depositions, inspections, witness interviews, expert conferences, trials, and many forms of discovery similarly contain an irreducible amount of human participation. AI may improve preparation for those events and reduce some surrounding work, but it cannot turn eight hours physically examining a witness into one hour.

The greatest pressure therefore falls on the categories of work that historically supplied enormous quantities of junior-associate hours: legal research, first drafts, document summaries, factual chronologies, comparisons, discovery drafting, internal memoranda, and document organization.

Those were also the activities that made traditional law-firm leverage economically powerful.

The Billable Hour Has Supposedly Been Dying Since 1999

All of this makes the current discussion about the “death of the billable hour” both more serious and, to someone who has practiced as long as I have, somewhat familiar.

I was reading articles predicting its demise on Law.com when I was a new associate in 1999. Corporate clients supposedly were no longer going to tolerate it. Alternative fee arrangements supposedly were about to replace hourly billing. Technology supposedly would permit firms to price legal work according to value rather than time.

Many of those arguments were good arguments. They remain good arguments. The prediction nevertheless proved premature.

The reason, I think, is that there is no pricing mechanism that eliminates the underlying importance of labor. A plumber may quote a flat fee, but the quote necessarily reflects an estimate of the labor, materials, overhead, and time required. The same is true of an electrician, HVAC contractor, mechanic, engineer, architect, tree service, dentist, or virtually any other provider of skilled services.

A flat fee does not make time disappear as an economic input. It reallocates the risk of estimating the time correctly.

This becomes particularly difficult in litigation because the lawyer does not control much of the workload. Opposing counsel may serve unexpected discovery. A witness may require a deposition. A judge may order additional briefing. An expert may raise a new issue. A case may settle in three months or proceed through trial and appeal over several years. No responsible lawyer can predict all of that precisely when the engagement begins.

That is one reason hourly billing has survived despite decades of predictions to the contrary. It is imperfect, but in genuinely unpredictable litigation it remains a rational way to allocate the risk of uncertain professional labor.

Why Clients Experience Legal Bills Differently

None of this means clients are irrational when they resist legal fees.

One reason legal bills are unusually painful is that people often need lawyers precisely because something has already gone wrong. A business believes another party breached a contract. A condominium association faces a serious construction defect. An individual has suffered a financial loss. A company has been sued. The client is therefore being asked to spend additional money in response to a problem the client frequently believes somebody else created.

The timing also works against the lawyer. Legal expense is immediate and certain; legal recovery is often delayed and uncertain. A client may have to write a $5,000 or $10,000 check today in pursuit of a recovery that may not occur for three or five years. Even a strong claim may ultimately fail, and no responsible litigator can promise otherwise.

That creates a very different psychological experience from many ordinary purchases.

I am reminded of this in my own life. Like anyone who owns a home, vehicles, bicycles, and other things that require maintenance, I routinely pay skilled people hundreds or thousands of dollars to solve problems. Over a relatively short period, I have paid substantial amounts to bicycle mechanics, a dentist, a plumber, an air-conditioning technician, a pool technician, and a tree service. Some individual visits cost several hundred dollars; more substantial work has cost more than a thousand or several thousand dollars.

I do not enjoy writing those checks. But I understand that skilled work has a price. The plumber has equipment, knowledge, overhead, and experience. The mechanic knows how to diagnose and repair something I cannot efficiently repair myself. The dentist spent years acquiring the knowledge and credentials necessary to perform the work safely.

Legal work is economically similar in that respect, although the uncertainty of the outcome makes the expense harder for clients to accept. A complex commercial case may require twenty, fifty, one hundred, or hundreds of hours of professional work. The fact that the client would prefer not to spend the money cannot change how much competent work the matter actually requires.

Medicine Illustrates the Problem of Professional Work Without Guaranteed Outcomes

The comparison with medicine is particularly useful because doctors, like lawyers, perform highly skilled professional services without guaranteeing outcomes.

A surgeon does not promise that surgery will cure the patient. A physician can make an appropriate diagnosis, recommend a reasonable treatment, perform competently, and nevertheless fail to produce the result everyone wanted. Human biology does not become predictable merely because a highly trained professional is involved.

The physician is still compensated for the professional service.

We accept this because the thing being purchased is not a guaranteed biological outcome. The patient is purchasing the physician’s training, skill, judgment, time, and adherence to the applicable professional standard.

Litigation is similar. A competent trial lawyer can identify the controlling law, preserve deadlines, develop the evidence, retain appropriate experts, conduct effective discovery, make strong arguments, and prepare a case properly. The lawyer nevertheless cannot control the judge, jury, witnesses, opposing counsel, or appellate court.

The principal economic difference is that medical expenses frequently are mediated through third-party payors. Employer-sponsored insurance, government programs, and private insurance arrangements mean that patients often do not experience the full professional cost of every medical service as a direct contemporaneous payment from their bank accounts. Whatever one thinks of the broader economics of health insurance, that financing structure materially changes the way consumers experience professional fees.

A plaintiff financing civil litigation often has no equivalent intermediary. The client writes the check directly while waiting years for a possible recovery.

That matters enormously.

Clients Are Not Always Cheap; Sometimes They Simply Do Not Have the Money

It is therefore too simplistic to say that clients who resist legal fees are merely cheap.

Sometimes they are. Every lawyer who has practiced long enough has encountered a client who demands extensive professional work while treating every invoice as evidence of betrayal. Lawyers are no different from other service providers in that respect.

But many clients face a real affordability problem. An individual with a valuable claim may have very little cash. A small company may have suffered the very financial loss that created the lawsuit. A condominium association may have substantial assets or reserves but also have competing obligations to owners, maintenance, repairs, insurance, and future capital expenses.

The fact that a claim may be worth $500,000 does not mean the claimant has $50,000 available today to finance litigation over it.

That is one of the central access-to-justice problems in civil litigation. A legally meritorious claim and an economically financeable claim are not necessarily the same thing.

Lawyers cannot solve that problem by simply working without compensation. They have employees, rent, software expenses, insurance, taxes, mortgages, families, and all of the ordinary economic obligations that their clients have. A law practice must generate revenue or it ceases to exist.

Nor can contingency fees solve every case.

Contingency Fees Transfer the Risk; They Do Not Eliminate It

Contingency representation performs an important function because it enables clients without sufficient liquidity to pursue claims. It also transfers an enormous amount of financial risk from the client to the lawyer.

A contingent-fee lawyer may work for three, four, or five years without being paid for professional time. The lawyer may carry substantial costs as well. A successful trial may be followed by post-trial proceedings and an appeal, delaying compensation further. If the case ultimately produces no recovery, the lawyer may receive nothing for years of labor.

The model can be highly profitable when practiced successfully across a portfolio of cases, as the largest plaintiffs’ firms demonstrate. But the economics depend on successful cases compensating the firm for unsuccessful cases and for the time value and risk of capital tied up for years.

There is no obvious reason why every civil lawyer should be required to become an investor in every client’s claim.

An hourly lawyer makes a different bargain: the lawyer agrees to provide professional labor and judgment at an agreed rate, and the client agrees to pay for the work performed regardless of whether an inherently uncertain litigation process ultimately produces the desired result.

There is nothing inherently unethical or unfair about that arrangement.

Indeed, it closely resembles the way most professional services are purchased.

What Clients Are Actually Buying From a Good Lawyer

The discussion becomes distorted when professional value is reduced entirely to hours.

Clients are not really purchasing time for its own sake. Nobody hires a lawyer because the objective is to own fifty units of lawyer labor. The client wants a contract enforced, a lawsuit defended, damages recovered, insurance coverage established, a transaction completed, a regulatory problem solved, or a legal risk understood.

Hours are the mechanism by which the professional labor is priced.

What a good lawyer actually supplies is a combination of knowledge, judgment, execution, and responsibility. The lawyer must identify the correct legal issues, locate and understand the governing authorities, recognize the evidence necessary to prove the client’s position, preserve procedural rights, anticipate the adversary, and make decisions under uncertainty.

Experience matters because much of that work cannot be reduced to rules mechanically applied. After decades of practice, the value of an experienced litigator lies partly in recognizing what not to do: which theory is a distraction, which discovery is unlikely to justify its cost, which apparent weakness is actually immaterial, and which apparently minor problem could destroy the case if not addressed early.

This is where AI can amplify rather than replace experience. Technology can provide the lawyer with more information more quickly. The experienced lawyer can then spend more time deciding which information matters.

How I Use AI in My Own Practice

My approach to this technological change is relatively straightforward. I use AI extensively, and I bill clients for the time I actually work.

If technology performs part of the preliminary research and organizational work that would have required many hours from a junior associate earlier in my career, I do not bill the client for hypothetical associate hours that nobody worked. The client receives that efficiency.

I still perform the senior-lawyer work. I read the authorities. I verify cases through conventional legal research and primary sources. I review the governing contracts and policies. I study the factual record. I determine whether the proposed analysis is correct. I identify contrary arguments, make strategic decisions, counsel the client, protect deadlines, draft and argue motions, conduct discovery, take depositions, and, when necessary, try cases.

Consequently, a client may receive an invoice reflecting six or seven hours of my actual professional time for an assignment that under a traditional multilayer staffing model could have generated substantially more total attorney time. That does not mean the six or seven hours are too many. It may mean the client never had to pay for the substantial first-stage associate work that historically preceded them.

For a sophisticated small firm, that can be an important competitive advantage.

Small Firms Can Compete Differently in the AI Era

Large law firms historically possessed an enormous infrastructure advantage. They had many associates, substantial libraries, research departments, litigation-support staff, document-management systems, and specialists in numerous adjacent fields. A solo lawyer could possess substantial substantive and courtroom experience and still lack the capacity to process information at comparable speed.

Modern technology has narrowed that gap considerably.

Research platforms, electronic discovery, cloud systems, digital filing, document automation, and generative AI now allow experienced small-firm lawyers to perform certain kinds of work that once required significantly more personnel. There remain enormous matters that legitimately require enormous teams. No technology turns one lawyer into fifty lawyers.

But much commercial litigation does not require fifty lawyers. It requires a lawyer capable of identifying the governing law, understanding the evidence, preserving deadlines, working effectively with experts, conducting discovery, writing persuasive motions, negotiating intelligently, and trying the case if necessary.

When a technologically sophisticated small firm can perform those functions competently, the client may obtain senior-level legal work without financing the entire traditional staffing pyramid.

That is a meaningful economic development.

The question should not be whether five lawyers appear on the invoice or whether the firm occupies several floors of an office tower. The relevant questions are whether counsel can handle the matter competently, whether the lawyer exercises good judgment, whether the necessary work actually gets done, and whether the total cost makes sense in relation to what is at stake.

Efficiency Has to Benefit Both Clients and Lawyers

There remains a difficult economic question. If technology permits a lawyer to produce in three hours work that once required fifteen, should the entire financial benefit belong to the client?

Under pure hourly billing, it largely does. The lawyer records three hours instead of fifteen.

That is excellent for the client, but it produces a strange incentive for the lawyer. The more technologically capable and efficient the lawyer becomes, the fewer hours are available to sell. A lawyer who invests time and money becoming dramatically more productive may actually reduce revenue unless the saved time can immediately be replaced with additional work.

That is why alternative fee arrangements continue to deserve consideration even though predictions about the death of hourly billing have repeatedly proved premature. Flat fees can reward efficiency while giving clients price certainty. Hybrid arrangements can divide risk. Contingency fees can align compensation with recovery where the lawyer is willing and financially able to assume the risk.

Different matters justify different arrangements.

What should not happen is fictional hourly billing. If AI reduces a ten-hour task to three actual hours, a lawyer charging by the hour should not reconstruct the vanished seven hours. The profession’s ethics authorities have made that clear.

But clients likewise should recognize that the three hours of senior professional judgment did not become worthless merely because technology eliminated seven hours of preliminary labor.

The Profession Should Defend Good Lawyering, Not Hours for Their Own Sake

There is a danger that the current enthusiasm surrounding artificial intelligence will produce another unrealistic expectation: that sophisticated legal work should suddenly become nearly free.

It will not.

Good lawyering is difficult work. It requires continuous learning because the law changes, facts differ, clients present new problems, and adversaries develop arguments that must be answered. Lawyers are asked to make consequential judgments based on incomplete information and then defend those judgments before judges, juries, regulators, opposing lawyers, clients, and sometimes appellate courts.

There is no point in arguing whether that work is “harder” than medicine, engineering, accounting, or another learned profession. Each discipline has its own demands and responsibilities. The relevant point is that law is a genuine profession requiring years of education followed by decades of continued learning, and sophisticated litigation does not become simple merely because the lawyer has practiced for a long time.

In some respects, experience makes a lawyer more aware of complexity rather than less aware of it. The longer one practices, the more examples one has seen of apparently easy cases becoming difficult, seemingly dispositive facts proving irrelevant, witnesses changing testimony, appellate courts distinguishing precedents, and litigation taking unexpected turns.

Clients are entitled to efficiency. They are entitled to honest billing. They are entitled to ask whether a proposed expenditure makes economic sense. They are entitled to terminate a representation when the likely recovery no longer justifies the cost.

But they are not entitled to pretend that difficult professional work requires no time merely because technology makes part of the process faster.

Artificial Intelligence Did Not Kill the Billable Hour; It Exposed What the Hour Was Measuring

The billable hour was never a perfect measure of legal value. It measured one important input: professional time.

Artificial intelligence is reducing the amount of that input required for some tasks, particularly the junior-level information gathering, organization, and initial drafting that once consumed large portions of law-firm workdays. That change is putting pressure on traditional associate quotas and on law-firm pyramids built around selling large volumes of junior time.

At the same time, AI is not eliminating the parts of legal work that require verification, judgment, advocacy, client counseling, procedural responsibility, and real-time participation in contested proceedings.

The profession therefore should resist two opposite mistakes. Law firms should not preserve old revenue by billing phantom hours for labor that technology eliminated. Clients should not conclude that because the preliminary work became faster, the experienced lawyer’s remaining professional work somehow ceased to have value.

Young lawyers should not be forced to choose between using technology efficiently and satisfying hourly quotas developed for an earlier era. Experienced lawyers should not be embarrassed to charge reasonable fees for the time genuinely required to read, verify, analyze, and advise. Clients should receive the economic benefit of technology without expecting lawyers to operate as uncompensated financiers of every legal dispute.

I have been reading predictions about the death of the billable hour since 1999. I am not prepared to predict its death now.

What I am prepared to predict is that artificial intelligence will continue changing what a legitimate billable hour contains. It will remove some work, concentrate other work at a higher level of judgment, and make it increasingly difficult to justify business models based primarily on maximizing the quantity of time consumed.

That may ultimately be healthy for the profession.

A good lawyer should not be valuable because the lawyer can generate a large number of hours. A good lawyer should be valuable because the lawyer knows what has to be done, does it correctly and efficiently, tells the client the truth about the risks, and is capable of carrying the matter through when the easy answer does not work.

Artificial intelligence can help with that work. It cannot eliminate the need for it.