There is a question crossing practically all professions:
“Will AI replace me?”
Lawyers ask.
Designers ask.
Programmers, teachers, journalists, doctors, advertisers, and analysts do too.
The question seems reasonable because it makes us imagine a very clear moment.
Today there is a person in that chair.
Tomorrow there is a machine.
As long as that day doesn't arrive, we keep working and interpret the permanence of our position as a sign of security.
But it probably won't be like that.
Your profession may continue to exist.
Your job title may keep the same name.
Your company may continue to hire people for the same area.
And yet, everything that gave security, bargaining power, and a future to that job may begin to change.
The risk doesn't start when the profession disappears. It starts when it continues to exist, but starts needing fewer people.
A profession is not just one thing
Imagine a young marketing analyst named Clara.
Clara is starting her career.
During the week, she researches competitors, organizes information, prepares presentations, writes first drafts, monitors campaigns, and transforms results into reports.
None of these tasks, in isolation, defines what it means to work in marketing.
But together, they occupy a large part of her routine.
They also teach Clara how to practice the profession.
When she researches a competitor, she learns to distinguish an important change from an irrelevant novelty.
When she prepares a report, she learns which numbers look good and which actually represent results.
When she writes a first draft, she understands how an idea changes according to the audience, the product, and the moment.
When she follows someone more experienced reviewing her work, she begins to perceive criteria that she couldn't yet explain on her own.
Now imagine that Clara's company adopts AI.
Initial research takes minutes.
Data arrives organized.
The presentation automatically gets a first structure.
The report already appears with a summary.
Ten versions of a text can be produced before Clara finishes the first one.
Did the marketing profession disappear?
No.
Did Clara lose her job?
Not necessarily.
But something important happened.
A profession is not a single block.
It is formed by tasks, decisions, knowledge, relationships, responsibilities, judgments, and assumed risks.
AI doesn't need to master all of this to transform the market.
It just needs to enter the right parts.
The profession is one of the last things to disappear. Tasks change first.
The first thing that disappears might be the ladder
It's easy to look at Clara's initial tasks and see only operational work.
Research.
Organization.
First drafts.
Reports.
It seems great that a machine can take this weight off her routine.
Except those tasks were also the first rungs of her training.
Clara didn't produce reports just to deliver reports.
She was learning to look at a business.
She didn't research competitors just to fill a presentation.
She was building a repertoire.
She didn't write imperfect versions just because someone needed to do them.
She was developing the judgment that, years later, would allow her to recognize a good version.
This mechanism appears in many professions.
A junior lawyer learns by researching documents, comparing decisions, and preparing first drafts.
A programmer starts by fixing minor problems, writing simple parts, and trying to understand systems built by others.
A designer develops aesthetic direction by producing variations, adapting pieces, and receiving criticism.
An advertising professional learns by researching references, testing approaches, and observing why some ideas survive review and others don't.
When AI takes over these tasks, the company doesn't just save time.
It may stop needing as many people at the beginning of their careers.
An experienced professional, accompanied by AI, starts delivering what previously required several beginners.
The company opens fewer positions.
Fewer people enter.
Fewer people accumulate experience.
The career loses some of its first rungs.
This is already appearing in the first available data.
A study by the Stanford Digital Economy Lab, based on payroll records in the United States, found a relative 16% drop in employment for workers between 22 and 25 years old in occupations most exposed to AI, after considering differences between companies.
In the same study, more experienced professionals in the same occupations were relatively more protected.
The authors themselves warn that the study is observational and does not prove that all the difference was caused by AI.
Even so, the pattern deserves attention.
The initial pressure doesn't seem to hit everyone in the same way.
It can start precisely with those who still needed to enter to learn.
AI may not eliminate a profession. It may eliminate the way people learned to practice it.
And that creates an uncomfortable question:
If the jobs that formed experienced professionals disappear, where will the next experienced professionals come from?
The position remains, but the value migrates
Clara stays at the company.
Her job title continues to be marketing analyst.
On the outside, everything seems relatively normal.
On the inside, the bar has changed.
Before, the company expected her to research, organize, and produce.
Now, as these steps have become faster, it expects Clara to deliver more.
It's not enough to present ten ideas.
She needs to recognize which two make sense.
It's not enough to generate a report.
She needs to explain what should be done based on it.
It's not enough to produce a campaign.
She needs to understand the audience, identify risks, defend choices, and be responsible for the results.
The productivity gain doesn't automatically turn into free time.
Often, it turns into a new expectation.
One person starts delivering what previously required several.
Teams get smaller.
The amount of production increases.
Execution becomes abundant.
And what becomes abundant usually loses economic value.
Maybe AI won't replace all designers.
But it can allow one great designer to produce what previously required a team.
Maybe it won't replace all lawyers.
But it can reduce the number of people needed to research, organize, and prepare a case.
Maybe it won't replace all marketing professionals.
But it can make the ability to just produce texts, presentations, and reports insufficient.
The profession remains in the market.
The number of available spots within it is what changes.
And value begins to concentrate in those who can do something beyond execution.
When a task becomes abundant, knowing how to execute it is no longer enough to guarantee value.
If executing stops being rare, what remains valuable?
The same technology that breaks the ladder can help build another
Here there is an important contradiction.
AI reduces the need for some tasks through which beginners learned.
But it can also give a person access to an amount of practice, knowledge, and capacity that would previously have been impossible to obtain alone.
This doesn't happen automatically.
Clara can use AI in two completely different ways.
In the first, she sends a task, receives an answer, copies the result, and delivers it.
She seems productive.
But she doesn't know how to explain why that answer is good.
She doesn't recognize errors with certainty.
She doesn't develop her own criteria.
The more she produces, the more dependent she becomes on something she doesn't know how to evaluate.
She is using AI to avoid work.
But she is also avoiding part of the learning.
In the second way, Clara uses AI to expand her training.
She asks the tool to explain different strategies.
Compares alternatives.
Tries to make a decision before consulting the answer.
Asks for criticism of her own reasoning.
Simulates scenarios.
Studies previous campaigns.
Records her mistakes.
Creates her own projects.
Tests hypotheses she wouldn't have the time or resources to test alone.
Instead of asking for just one answer, she builds a cycle:
try, compare, receive feedback, correct, and try again.
In this case, AI doesn't replace practice.
It increases the amount and speed of possible practice.
But there is an important limit.
AI doesn't replace real problems.
It doesn't replace real consequences.
It doesn't replace mentorship, contact with more experienced people, external feedback, and progressive responsibility.
A tool capable of answering anything can also confirm a bad idea very convincingly.
Therefore, the new ladder will not be built only with prompts, agents, and automations.
It will need to combine AI with real projects, human criticism, study, experimentation, and responsibility.
The difference will be between using AI to avoid learning and using it to accelerate the construction of judgment.
The technology that removes some of the old rungs can also help create new ones.
Except this time, part of the responsibility for building that ladder may leave the company and fall on the professional themselves.
The birth of the augmented professional
Over time, Clara stops using AI only when she needs to finish a task.
She begins to build a capacity around herself.
Organizes a memory of previous campaigns.
Records decisions, results, and errors.
Creates a recurring process to research competitors.
Defines criteria for evaluating ideas.
Configures agents for specific tasks.
Automates follow-ups.
Keeps the context of each client organized.
Creates verification steps before any delivery.
Clara no longer starts each project from scratch.
She carries with her everything she has learned and a system capable of putting that knowledge into motion.
This is a bigger difference than simply knowing how to use ChatGPT.
There is the professional who uses AI to produce the same work faster.
And there is the professional who transforms AI into memory, process, repertoire, and accumulated capacity.
The first saves time.
The second changes what a single person can take on.
This change already appears in the way large companies describe work.
In the 2025 Work Trend Index, Microsoft presented the idea of teams formed by humans and agents, with professionals directing “digital colleagues” to execute specific parts of the work.
The Anthropic Economic Index also found two distinct patterns of use: automation, when the task is delegated, and augmentation, when the person uses AI to learn, validate, and develop the work together.
The distinction matters because clicking a button does not create a lasting advantage.
Over time, almost everyone will have access to the same tools.
The advantage will be in the system built around them.
What information did you organize?
What context did you accumulate?
What criteria did you develop?
What processes do you know how to coordinate?
What results can you verify?
What responsibilities did you start to assume?
In the future, a company may not only evaluate a person's experience, training, and previous results.
It may also want to know what capacity that person can mobilize.
What processes can they operate?
How much context can they carry?
How do they control the quality of what their agents produce?
How much result can they generate without expanding the team?
The person doesn't just arrive with a resume.
They arrive with a kind of professional infrastructure of their own.
The company doesn't just hire the person. It hires their judgment and all the AI capacity they have learned to build and orchestrate.
Each professional can carry with them an operational capacity that previously belonged only to an entire company.
But that doesn't mean the machine automatically made the professional valuable.
The machine can generate.
Someone still needs to give direction.
The machine can suggest.
Someone still needs to judge.
The machine can process a huge amount of information.
Someone still needs to understand the context.
The machine can recommend an action.
Someone still needs to answer for the consequences.
Direction, judgment, context, and responsibility are not a romantic defense of the human being.
They are concrete parts of the work.
The cheaper production becomes, the more important they become.
Capacity without direction, judgment, context, and responsibility is not yet a complete profession.
The test you can do today
Don't wait for your profession's name to appear on a list of threatened jobs.
Take a normal work week and list what you actually do.
Don't just write “I'm a lawyer,” “I'm a designer,” or “I work in marketing.”
List the activities.
Then, classify each one into six groups.
1. Production
What do you create, organize, or execute?
2. Decision
What requires a choice of yours?
3. Context
What depends on knowing the company, the client, or the situation deeply?
4. Responsibility
What does someone need to answer for?
5. Trust
What depends on relationship, reputation, or credibility?
6. Learning
Which tasks are forming the experience necessary for you to take on larger jobs?
Now identify what AI can already start.
Don't just ask if it can do everything perfectly.
That's too comfortable a bar.
Ask:
Can AI already produce a useful first version?
If the answer is yes, it can already alter the time, cost, and number of people needed for that task.
Then, look for the invisible risk.
If this task loses value, what happens to you?
Is it just operational or is it also part of your learning?
If anyone can execute it with AI, why will someone continue to choose you?
Are you developing judgment or just increasing speed?
Finally, look for the invisible opportunity.
How can you use AI to practice more?
What knowledge do you need to organize?
What process can you build instead of just accelerating?
What criteria do you need to develop to evaluate the results?
What greater responsibility can you assume now that execution requires less time?
The most important question is not:
“How can I finish my task faster?”
It is:
“Am I using AI to do the same work faster or to become a professional with a different capacity?”
The risk starts earlier. The opportunity too.
AI probably won't arrive on a Monday to announce that your profession has ceased to exist.
Your job title may remain.
The company may continue to hire.
People may continue to practice the same activity for many years.
The change will happen in a less visible way.
First, some tasks will take less time.
Then, one person will be able to deliver what previously required several.
Entry-level positions will decrease.
The old ladder will lose rungs.
Expectations will increase.
Certain skills will stop being rare.
Value will migrate to new parts of the work.
When someone finally asks if the profession has been replaced, that may already be the wrong question.
The most vulnerable professional is not necessarily the one whose profession can be automated.
It is the one whose value depends only on tasks that are becoming easy to reproduce.
The most prepared professional doesn't try to prove they can work without AI.
They learn to direct the technology, verify its results, and assume responsibilities that the machine alone cannot carry.
This doesn't make anyone impossible to replace.
But it makes a prepared person much more capable, valuable, and difficult to replace than they would be working alone.
The future will not be divided only between people and machines.
It will be divided between people who continue working alone and people who have learned to transform machines into an extension of their own capacity.
AI is not going to replace your profession all at once. The real risk starts much earlier than that.
But so does the opportunity.
If you want to prepare for this change
The best way to prepare is not to compete with AI.
It is to learn to transform it into your own capacity.
If you want to understand how to do this in your work, follow me here on X.
I share daily practical ways to use new technologies to produce better, take on greater responsibilities, and become more valuable in the market.
Sources cited
- Erik Brynjolfsson, Bharat Chandar, and Ruyu Chen. Canaries in the Coal Mine? Six Facts about the Recent Employment Effects of Artificial Intelligence. Stanford Digital Economy Lab, November 2025 version.
- Microsoft. 2025 Work Trend Index Annual Report: The Year the Frontier Firm Is Born.
- Anthropic. Anthropic Economic Index: Uneven Geographic and Enterprise AI Adoption, September 2025.
