AI will not take your job. Someone who knows how to use it will

The line is true, but it gets used wrong. It has turned into a comfort: relax, the machine is not coming for you, you just have to learn to use it. It is not a comfort. It is a warning with a deadline, and the deadline has passed.

I am writing this in September 2026 because I want it on record when I thought it. For close to two years I have been putting artificial intelligence into production, not into slide decks, inside an agency with a real team and clients who pay. What I have seen does not look much like what gets told from a stage.

What the line actually means

Nobody is going to walk into your office and tell you a model is replacing you. That is not how it happens. What happens is slower and more uncomfortable.

What happens is that someone at your level, on your salary, starts delivering in two hours what takes you two days. Not because they are smarter, but because they spent three afternoons understanding a tool you still watch from a distance. What happens is that the work you used to do becomes an intermediate step in someone else’s process. What happens is that when a new project gets handed out, you are no longer the obvious choice.

In one sentence: AI does not fire you. It makes you comparable to someone delivering three times as much, and you lose that comparison.

What I have actually seen happen

I would rather describe what has been in front of me than what the reports say.

The boring work disappears first, and that is good. In my own agency there are tasks that two years ago consumed whole afternoons of qualified people and that nobody touches today. Nobody lost their job over it. What changed is where their day goes.

The people who resisted did not lose their jobs, they lost ground. That is subtler and worse. They are still in their roles, but they are no longer the ones asked first. It is a slow exit and almost nobody sees it while it is happening.

AI projects that fail almost never fail on the technology. They fail because nobody defined which problem was being solved, because the data was dirty, or because a platform was bought before there was a use case. The technical part is the part that breaks least.

And the big effect is not replacing people, it is doing what was not being done at all. A small team that can now analyse every one of its sales conversations instead of a sample did not lay anyone off: it started doing something that used to be impossible on cost. That is the real shift, and hardly anyone talks about it because it does not make headlines.

Why this plays out differently in Latin America

Across the talks I have given in Honduras, Colombia, Spain, Ecuador, Costa Rica, Aruba, Bonaire and Curaçao, the conversation is more alike between countries than you would expect. But there is one difference that matters.

In a large company in a wealthy country, adopting AI is a project with a budget, a committee and a consultancy. In a Latin American SME, adopting AI is one person with judgement and a free afternoon. That sounds like a disadvantage and it is exactly the opposite: the distance between deciding and having something running is far shorter here.

What holds the region’s SMEs back is not access to the technology, which now costs very little. It is that nobody inside the company has the permission, the time or the judgement to try. Buying software does not fix that.

What I would do this week if this worries me

Not a three-year plan. Four concrete things, in this order:

Pick a repetitive task you do every week and automate it yourself. Not the most important one: the most boring one. Writing the weekly email, summarising meetings, sorting requests. Give it two weeks of failed attempts. The point is not the task, it is that you learn where the tool helps and where it lies.

Learn to evaluate, not just to ask. The skill that gets expensive is not writing elegant instructions: it is recognising quickly when the output is wrong. You can only do that if you know the subject. Your experience does not lose value, it changes function: it moves from producing to judging.

Put AI where you have data, not where you have enthusiasm. Value shows up when the model works on information of yours that nobody else has. On public information, what you get is what your competitor gets.

Tell people what you are doing. It sounds like marketing and it is survival. If you automated something and nobody in your organisation knows, for practical purposes you did not.

What I do not believe

I do not believe everyone has to become technical. Most of the people I have seen get real value out of this do not write code: they know their craft well and learned to hand the mechanical part to the machine.

I do not believe this is free or painless. Some jobs will shrink, and saying otherwise to reassure people strikes me as dishonest.

And I do not believe the advantage of having started early lasts forever. Tools get cheaper and easier. What does not get cheaper is the judgement to know what is worth automating, and that is built by using them.

So what?

The line in the title has been repeated from stages for years, and I think it should be said in full: it is not a comfort, it is a deadline. The person who knows how to use it is not some future profile or distant expert. It is your colleague, and they are three afternoons ahead.

The good news is that three afternoons can be made up. The bad news is that in a year they will be thirty.

If you are thinking about bringing this conversation into your company or your event, that is what I speak about, with cases that are in production rather than on a slide.

Frequently asked questions

Is it true that AI will eliminate jobs? Some, especially those made of repetitive, well-defined tasks. But what I have seen most often is not the role disappearing: it is the same person starting to do different things, and whoever does not adapt losing relevance inside their own role before losing the role itself.

What should a non-technical person learn first? To evaluate output in their own field. Writing instructions takes a week to learn; recognising when an answer is wrong can only be done by someone who knows the subject. That is where your experience is worth more, not less.

Where does an SME with no technical team start? With a repetitive process they already run every week that costs measurable time. In most Latin American cases, that process is customer service over WhatsApp.

How much does it cost to start? Far less than people assume. The expensive part is not the tools: it is the time of someone with judgement to define what gets automated and to correct it during the first weeks.