The Efficiency Paradox: When AI Also Puts Your Profession at Risk

View on LinkedIn

We all want efficiency — until you are the next item on the list of inefficiencies.

I.

If you live or have lived in the United States, there are two things you know with absolute certainty: you do not want to go to the doctor, and you do not want to go to a lawyer.

Not because you do not need them. But because the price of needing them can throw you into financial disarray — leaving your credit cards overdrawn after an emergency room visit or a legal consultation that lasted a couple of hours.

A legal consultation starts at three hundred dollars an hour. An emergency room visit can cost as much as a month's rent. And the interesting part is not that they are expensive — there are reasons for that — but that the entire system is built with layers of mandatory requirements that restrict the average citizen even in the most straightforward matters.

Consider something as routine as a prenuptial agreement. Two people who agree on everything. Who want to put their terms in writing before getting married. In most countries, that is a formality. In the United States, the system requires each prospective spouse to retain their own attorney for the document to be valid. Each person pays at least fifteen hundred dollars in legal fees, or the agreement simply does not exist. It does not matter that both parties want exactly the same thing.

You need a prescription to buy antibiotics that sit over the counter at any pharmacy in Mexico City. You need a lawyer to understand a lease agreement that, in theory, you are signing voluntarily.

The system was sufficiently protected before artificial intelligence existed. What AI did was something no one anticipated.

It did not attack the system. It exposed it.

II.

Let us be honest. An immigration attorney who has processed hundreds of visas has a value that no machine can replicate. The intuition of someone who has seen the pattern. The judgment that forms after years in front of the same tribunal. The knowledge that this judge has this tendency, that this officer checks this first. I would not entrust my immigration case to a chatbot. You pay for the expertise of someone who has handled the process hundreds of times, and that has real value.

But there is an enormous difference between an immigration case that can affect your status and force you to dismantle your life, and understanding what an eviction notice says. Between surgery and googling at two in the morning whether what you are feeling warrants a trip to the emergency room.

For decades, both things — the simple and the complex — lived behind the same three-hundred-dollar-an-hour barrier. And then, for the first time, millions of people had access to something that resembled — imperfectly, but resembled — basic guidance. No appointment. No copay. No surprise bill three weeks later.

It was not perfect. But for someone who previously had nothing, it was everything.

The logical question was: how do we improve this?

Ban it.

III.

In 2023, an entrepreneur named Joshua Browder attempted something bold. His company, DoNotPay, proclaimed itself "the world's first robot lawyer." The promise was irresistible: a chatbot that could substitute for an attorney, generate valid legal documents, and even represent you in court via an earpiece connected to the AI. Instant legal access. No prohibitive costs.

It turned out to be a disaster.

The documents had errors. Browder never tested whether his chatbot performed at the level of a real attorney. He never retained lawyers to validate the quality of the product. They threatened to send him to prison if his AI spoke in a courtroom. In 2025, the FTC sanctioned him: a hundred and ninety-three thousand dollars in fines, a prohibition on making unsupported claims, and a mandatory notice to all subscribers that the service did not substitute for an attorney.

The system stopped him. Without a new law. Using the tools that already existed.

What happened next is what deserves careful examination.

Because the reaction was not to say "the regulatory framework worked, the charlatan was punished." The reaction was to propose laws that punish everyone. The charlatan and the one who provides accurate information alike. The irresponsible startup and the community clinic that uses AI to explain to a tenant what their rights are.

In February 2026, New York Senate Bill S7263 reached the Senate floor calendar. It is not yet law — it must pass a floor vote, cross to the Assembly, and be signed by the Governor — but it advanced through committee with a unanimous vote and has real momentum. The bill is barely two pages long. It prohibits any chatbot from providing "substantive responses, information, or advice" in fourteen licensed professions, plus the practice of law:

"A proprietor of a chatbot shall not permit such chatbot to provide any substantive response, information, or advice, or take any action which, if taken by a natural person, would constitute a crime under section 6512 or 6513 of the education law" — NY Senate Bill S7263, §390-f(2)(a)

The drafting is convoluted, but what it says is simple: no one who is not a licensed professional may guide you. Not even a machine.

It creates a private right of action with fee recovery. And the truly extraordinary detail: the operator cannot escape liability even if it warns the user that they are speaking with an AI. The disclaimer does not protect you.

In the same week — March 2026 — three thousand five hundred kilometers away, in San Luis Potosí, Mexico, a citizen submitted an initiative before the state legislature to add Article 259 BIS to the Penal Code. The proposal equates to the crime of unauthorized practice of a profession anyone who designs, develops, or commercializes AI systems that generate advice characteristic of a regulated profession without the involvement of an authorized professional. The penalty: one to four years in prison.

Two hemispheres. Two legal systems. The same reaction.

IV.

What is happening with lawyers in New York and San Luis Potosí is not new. It is what every professional guild does when it feels its interests are threatened: legislate.

Taxi drivers tried it with Uber. Hotels tried it with Airbnb. Record labels tried it with Napster. The pattern is always the same — first they ignore it, then they mock it, then they fight it in courts and legislatures.

Sometimes they buy time. They never win the war.

V.

And this is where the questions that no one wants to answer begin.

Because we all agree on doing what is most efficient for society — until you become the next item on the list of inefficiencies.

Then we ask for protection. For the graphic designer who gets fewer commissions because people now make their own logos with AI. For the translator who lost a quarter's worth of work overnight. For the taxi and Uber driver whose days behind the wheel are numbered. For the attorney who watches a chatbot do in thirty seconds what he charges an hour to do.

So do we want efficiency? Or do we want it only when it is someone else's turn? Where is the happy equilibrium?

VI.

We are living a historic moment, and we are watching it unfold in real time.

Jack Dorsey — co-founder of Twitter — publishes a letter to investors explaining that artificial intelligence has already changed the way his company Block operates. He eliminates more than four thousand jobs overnight. The stock goes up. Four thousand people find out through a post that fits on a phone screen.

Those graduating from universities find fewer and fewer opportunities. And those who do find them arrive in a professional world that is no longer what it was, but has not yet finished evolving. They are being tossed by a wave that no one taught them to surf.

Meanwhile, in Albany they draft laws so that a chatbot cannot explain a lease agreement to you. In San Luis Potosí they propose prison for whoever programs an AI that provides guidance on legal matters. And everywhere, consumers — who are also workers, drivers, designers, and recent graduates — keep opening ChatGPT at eleven at night because it is the only thing they have.

I am not going to conclude with what should or should not be done. I do not have that answer. What I can say is something simpler and more uncomfortable:

What they told us was going to happen — is already happening.

Official bill page (NY Senate): https://www.nysenate.gov/legislation/bills/2025/S7263

Full text in PDF (2 pages): https://legislation.nysenate.gov/pdf/bills/2025/S7263

Citizen initiative San Luis Potosí: https://drive.google.com/file/d/1HOGKYX7p5YEmaiwWtIRsUyKoz1eD6yqo/view