Johnson’s Law: The more confidently someone says, “AI will never replace X,” the more likely AI is to replace a frightening amount of X, and frighteningly fast.
Johnson’s Corollary: AI does not have to replace your profession to replace you. It only has to make someone or something else substantially more productive, less expensive, or better at the work than you.
Johnson’s Second Corollary: The parts of your job you are most certain AI cannot do are the parts you should examine first.
I call these Johnson’s Law and its corollaries partly—but only partly—as a joke. I am not predicting the extinction of lawyers, doctors, accountants, programmers, or other professions. A profession can survive while artificial intelligence changes how much human labor its work requires, what customers will pay for that labor, and which workers and business models endure. But survival is not to be confused with thriving.
I am a family law attorney, so most of my examples come from law. The mechanism is much broader: how much human labor does a useful result require, and what happens when technology changes that number?
Harvard Business School professor Karim Lakhani helped popularize the observation that “humans with machines will replace humans without machines.” In a 2023 Harvard Business Review IdeaCast interview, he added an asterisk worth remembering: “maybe we’ll need fewer of them.”
That describes Stage One. I am more interested in what comes next.
Stage One: The Adoption Advantage
AI does not need to make every professional more productive at every task or double anyone’s output. It only needs to reduce the labor required for enough valuable work to change the economics of the profession.
If, after necessary verification and professional review, a task that once required ten hours can reliably be completed in seven, that is already a substantial change.
I know where those gains come from in family law. AI can compare financial declarations, organize discovery responses, build chronologies from hundreds of communications, identify inconsistencies, summarize transcripts, compare proposed orders, prepare deposition outlines, and produce useful—often brilliant—first drafts of arguments based on a hearing record. The lawyer still needs—for now—to determine whether the result is accurate, legally sound, strategically sensible, and right for the particular case.
But a lot less human labor is required to get there.
At first, the professional who uses the technology well gains an advantage over the professional who does not.
Stage Two: The Productivity Reset
The more important stage begins when competent AI use stops being unusual.
Once nearly everyone in a profession uses comparable tools, merely using AI is no longer much of a competitive advantage. The productivity gain remains. If work that once required ten hours now routinely requires seven, customers, competitors, employers, and eventually the market start treating seven hours as normal.
For an hourly law firm, the pressure is immediate. A task that once legitimately generated ten billable hours now generates seven. The firm has three fewer hours to sell. It must find more work, reduce costs, change staffing, handle more matters, change how it charges, or accept lower revenue.
That is why “AI will never replace lawyers” tells us so little. Assume it is completely true. Assume lawyers will still appear in court, advise clients, negotiate settlements, exercise judgment, and take responsibility for their work—which, frankly, is not the most solid assumption. Even then, substantially fewer human hours may be required to provide the same amount of legal service.
AI does not have to eliminate the profession. It only has to change what the profession still needs humans to do.
AI and Robots Are Coming for Most of Our Jobs
AI and robots are coming for most of our jobs. They need not eliminate every profession or every worker. They only need to eliminate enough human labor that fewer people are required to produce what society wants.
Yes, lower costs can create additional demand, and some professions may even grow. But if one lawyer, accountant, programmer, analyst, physician, or other professional can eventually produce what once required several people, demand will have to grow enormously just to preserve the same number of human jobs. I do not expect it to.
The same principle will increasingly apply to physical labor. As robotics improves at seeing, moving, manipulating objects, navigating unpredictable environments, and following complex instructions, the supposed safety of work that requires a human body will erode too.
The effects will not be evenly distributed. Some people will become more valuable because technology multiplies what they can accomplish. Others will discover that the work on which their livelihood depends can now be produced faster, cheaper, and better without them at all. Entire professions need not disappear for enormous numbers of jobs within them to disappear.
The larger change is simple: far less human labor will be required to produce much of what society wants. That will dramatically increase what can be produced with a given amount of human labor. It will also be economically brutal for many people whose livelihoods depend on selling work that machines can increasingly do—and do better than humans can.
AI Is Already Better at a Lot of Law Than We Are
AI is already better at a great deal of law than I am—and I am no slouch as a lawyer. I take no pleasure in saying that, but I have no trouble admitting it. The rest of the legal profession needs to reckon honestly with the same fact.
AI can already read, compare, organize, summarize, synthesize, draft, revise, and spot issues across quantities of information no lawyer can process as quickly. It can hold more material in view at once, generate alternatives almost instantly, expose inconsistencies, and help test arguments without fatigue, boredom, or ego. In many parts of legal work, that is not merely “useful assistance.” It is superior performance.
Lawyers understandably retreat to “human judgment” as the last safe category. Sometimes that refuge is real. Reading a client, understanding a judge, knowing when not to make the argument you could make, recognizing when technically correct advice will produce a disastrous practical result, and taking responsibility for a decision are not the same thing as summarizing a record or drafting an argument.
But we do not need nearly as many lawyers to perform those functions as we currently need to perform the entire bundle of legal work. When—not if—AI takes over more of the research, analysis, comparison, drafting, organization, and routine problem-solving while humans retain the hardest judgment calls, that does not save the existing lawyer workforce. It shrinks the portion of legal work for which a human lawyer is actually needed.
Some human judgment may remain indispensable. That does not mean the millions of human hours currently sold as legal services will remain indispensable too.
And we should stop using “judgment” as a magic word that protects everything lawyers currently do. Much of what we have historically called legal judgment is retrieval, comparison, pattern recognition, synthesis, and applying known criteria to facts. AI is already very good at those things and getting better.
That leaves a harder question than whether AI can “practice law.” Harder only because the answer is hard to take, not hard to find. Yes, AI can already do essentially what we now define as practicing law. It performs large portions of legal work astonishingly well and is only getting better at it.
The real question is which parts of being a lawyer remain distinctly and economically valuable once machines do much of the intellectual work better, faster, and cheaper than we do. Professional pride has no bearing on the answer.
Do Not Ask Only How Technology Makes Today’s Work Faster or Cheaper or Better
There is a still larger question: not only how technology can help us do today’s work faster, cheaper, or better, but how tomorrow’s problems will be solved—and whether solving them will still require the same people, professions, tasks, and business models that solve them today.
Technology can do more than improve the traditional service. It can change the form of the service, who provides it, how much human involvement it requires, and whether some traditional work needs to be performed at all.
Lawyers naturally imagine the future of law by starting with the lawyer and asking how technology will make the lawyer faster, cheaper, or better. Doctors can make the same mistake about medicine, accountants about accounting, programmers about software, and teachers about education.
The people who need the service generally care less about preserving the traditional professional role than about getting the underlying problem solved well, affordably, and with as little unnecessary friction as possible.
That changes the question. The threat to a profession may not be that AI learns to perform every traditional task within it. It may be that technology enables people to solve some of the problems for which the profession was historically the best available answer without buying the traditional professional service at all.
“AI can never do everything I do” is not much comfort if nobody still needs everything you do.
After Everyone Has AI
The first AI question was whether machines would replace professionals. They will. They already are. The soothing answer was that professionals using AI would replace professionals who did not. But widespread adoption changes the question again.
Less human labor will be required to produce the same results. Most tasks will stop supporting the fees they once did. Most professional career paths, as we know them today, will disappear. Most problems will no longer be solved through the traditional professional service at all. And many things we confidently call “human judgment” will turn out to be work machines can do better than we can.
I could hedge and say I do not know whether AI will reduce the total number of lawyers. That would be evasive. AI will slash the number of lawyers needed across virtually every area of practice. It is happening already.
I could also comfort myself with the possibility that cheaper legal services will expand demand enough to absorb all that new capacity. That won’t happen either. There is substantial unmet demand for legal help, and lower costs will bring some of it into the market. But that added demand will not come close to offsetting the reduction in human labor AI will make possible.
A profession can—can—remain necessary while requiring far fewer members. The work can continue while much of the labor disappears. The market can grow while the number of people needed to serve it falls.
Your profession does not have to disappear for your job to disappear with it.
That is Johnson’s Law, such as it is: AI doesn’t have to replace your profession to replace you.
Utah Family Law, LC | divorceutah.com | 801-466-9277