By The Briefcase Team | June 3, 2026

A decade ago, every conference panel, every magazine cover, and every well-paid futurist with a TED Talk told us that AI would replace radiologists. Geoffrey Hinton, the man who basically invented modern AI, said in 2016 that we should stop training radiologists because they would all be unemployed by 2021.

It is 2026. Radiologists make more than $500,000 a year. Employment in the field is still growing. The number of CT, MRI, and PET scans performed in the U.S. has roughly doubled because AI has made each scan cheaper to read, and when something gets cheaper, people buy more of it.

This is not magic. This is a 160-year-old economic principle that British economist William Stanley Jevons figured out in 1865 while watching coal-powered steam engines. Jevons called it: when something becomes more efficient, total consumption of it goes up, not down.

The same thing is about to happen to real estate. Briefcase is here to tell you to relax. Mostly.

In 1865, James Watt's steam engine made coal three times more efficient. The conventional wisdom of the day was that Britain would burn less coal. Jevons published a book called The Coal Question, explaining that the opposite would happen: cheaper coal use would create a much larger market for coal-powered everything, and Britain would burn more coal dramatically.

Britain proceeded to burn more coal dramatically. Jevons was right. Everyone else updated their priors. The principle has held up across every major productivity shift since: cheaper light means more light, cheaper computation means more computation, cheaper communication means more communication.

Slok's April 2026 framing is that this is now happening to professional services. Cheaper legal work means more legal work. Cheaper accounting means more accounting. The total addressable market expands faster than the cost-per-task falls, and the number of people employed in the field grows.

The data backs him up. U.S. business formation is at the highest level in recorded history, young-worker unemployment is falling faster than older-worker unemployment, and entire categories of "this will be automated" jobs (translation, customer support, document review) have actually grown in headcount over the last 36 months.

Hinton's mistake in 2016 was not technical. The AI did exactly what he predicted. It got really, really good at reading scans.

His mistake was definitional. He thought "radiologist" was a job. It is not. It is a basket of tasks. Reading scans is one task in the basket. The other tasks (clinical correlation, multidisciplinary case conferences, procedural radiology, supervising AI outputs, talking to patients, and owning the legal accountability for a misread) did not get automated. They got augmented.

The radiologist who used to read 60 scans a day now reads 200. The marginal cost of imaging dropped 70 percent. Doctors order vastly more scans because they are cheaper. The total number of scans exploded. The total number of radiologists also grew. The salary went up.

Reading scans got cheap. Being a radiologist became more valuable. This is the pattern.

Here is where Briefcase puts on its reading glasses and gets serious for a minute.

Three real-estate-adjacent professions are about to live the radiologist paradox. Two of them will win bigger than they expect. One of them is going to feel like the lower-tier mortgage underwriter pre-Gateless. Honesty matters.

The National Association of Realtors had 1,509,195 members as of April 2026, per NAR Treasurer Greg Hrabcak. That is down about 2 percent year over year, which is causing the trade press to write breathless "the end of realtors" pieces.

It is not the end of realtors. It is the cycle. NAR membership swelled in 2021 and 2022 when anyone with a pulse could pass the licensing exam and close two transactions on commission. The 2026 number is what happens when the tide goes out: weak agents leave, strong agents stay, and the average productivity per member rises.

What AI will do to real estate agents is what AI did to radiologists. The boring tasks (drafting MLS descriptions, comparable analysis, follow-up scheduling, document gathering, and market reports for clients) cost an order of magnitude less. The hard tasks (negotiation, judgment calls on a specific property, handling a buyer's emotional roller coaster, sitting in a room with a seller who has just lost their spouse) do not get automated. They get more valuable.

A good agent in 2030 will close two or three times as many transactions per year as they did in 2024 because AI eats the drudgery. Total transaction volume rises as friction declines. The agent who learned the new tools wins. The agent who refused to learn them, the one who is still emailing Excel sheets, finds a new line of work.

This is not theoretical. BLS data shows real estate employment at 1.85 million workers as of April 2026, basically flat year over year despite a brutal transaction environment. The industry is consolidating around productivity, not collapsing.

Cheaper underwriting means cheaper loans, which means more loans. The same Jevons effect that doubled the number of CT scans will multiply the number of mortgage transactions, refinances, and home-equity products that get originated in the next decade. The platform exists. The cost-per-application is falling. The market is now open to anyone who has ever wondered if they could refinance.

This is great news for mortgage brokers, who own the borrower relationship, hold the NMLS license, and structure the product. JobZone scores loan officers at 29.8 in the "Yellow Urgent" zone, meaning meaningful transformation but not displacement. They keep their job. They originate three times the volume. The honest read is that the top quartile of mortgage brokers in 2030 will look more like investment advisors: relationship-led, judgment-led, with AI doing the document processing in the background.

Here is where the Jevons paradox does not save you, and Briefcase will not pretend otherwise. Mid-level mortgage underwriters processing conforming conventional loans through Desktop Underwriter score 20.9 on the JobZone risk index, firmly in the Red Zone. Gateless Smart Underwrite is already auto-clearing 70-75% of conforming files and targeting 85% by late 2026.

That job, specifically, the volume-driven conforming-loan underwriter, is shrinking. Not vanishing, but consolidating. A team of six in 2024 is reduced to three by 2028, handling the same volume.

But, and this is the radiologist paradox at work, the underwriting profession overall is not shrinking at the same rate. The work moves up the complexity ladder. Non-QM, jumbo, construction lending, self-employed income files, fraud-pattern detection, exception handling. The underwriter who learns the AI tools and becomes the exception specialist makes more money in 2030 than they do today. The underwriter who insists on doing what AUS already does is the radiologist who refused to use a CT scanner in 1985.

This is the cleanest version of how markets work. A cheap input creates a larger market, which in turn drives more total economic activity, which in turn creates more jobs in the same field at higher complexity and higher pay. The losers are the people doing the parts of the job that the machine does better. The winners are everyone else in the same industry.

Real estate, mortgage, and the entire housing stack is about to get the radiologist treatment. The total number of transactions goes up. The friction-per-transaction goes down. The job description changes. The headcount for high-value tasks is growing.

The lesson is not "AI will not take your job." The lesson is "the task you currently do may go away, but the profession you belong to will be larger, more lucrative, and more important in five years than it is today, if you reposition."

One. Audit your task list. Which 30 percent of your weekly work is comp pulls, follow-up emails, listing descriptions, document chasing, scheduling, and templated client updates? That is the part AI eats. The part you should be doing more of (judgment calls, relationship work, complex negotiations, market expertise) is the part that gets more valuable.

Two. Learn one tool deeply. Not five tools shallowly. Whether it is your MLS's AI features, a transaction coordinator AI, or a deal-analysis platform, become the person in your office who is fluent. That alone moves you from the displaced quartile to the augmented quartile.

Three. Read Slok (and Briefcase!!). The Apollo daily spark is free, it is short, and the man is the best macroeconomist in finance at translating data into useful framing. He has been right about more of the post-2020 cycle than anyone we read.

A 160-year-old British economist watching steam engines figured out the answer to "will AI take my job" before most of our great-grandparents were born. The answer is: it will take some of your tasks. The profession you belong to will grow, become busier, and pay better. The people who reposition early will eat the people who insist nothing will change.

The radiologists who were supposed to be unemployed by 2021 now make half a million dollars a year. The realtors, mortgage brokers, appraisers, and property managers reading this in 2026 will, if they play this right, look back on the panic of the mid-2020s the same way radiologists look back on Hinton's 2016 prediction. Quaintly. From a paid-off house.

TOP STORY

Text