Artificial Intelligence and Its Impact on the Workplace

What is artificial intelligence, or AI? Why does everyone suddenly talk about it? Is it true it’s going to end many people’s jobs? It’s natural for humans to feel some fear in the face of change — a small sense of threat that can affect our perceived well-being (that’s at least what we talk about in my Change Management classes). Still, it’s worth clearing away some of the hype and reflecting on the future.

Fundamental Concepts

Before we start debating, it’s worth agreeing on what a few things actually mean. Here are some important definitions and concepts.

Artificial Intelligence — AI

The simulation of human intelligence in machines that can perform tasks, learn from experience, and adapt to new situations.

Machine Learning

The development and use of algorithms and statistical models to carry out tasks autonomously — that is, without explicit instructions. In simple terms, it’s the difference between hiring a kitchen apprentice you have to hand a recipe and detailed steps to, versus hiring a service you just ask for “something to eat,” and, with no further input from you, it hands you a great meal.

Deep Learning

The use of ANNs — artificial neural networks, which simulate human neural networks with computing systems — to build machine learning systems. If you think about it for a second, the Internet is an interconnected network of computing equipment that could itself be used as a neural network — simulating the behavior of a complex system with hundreds, thousands, or millions of interconnected machines.

Natural Language Processing — NLP

The ability to decode and encode natural language — the language humans use to communicate, with spoken/written language being a subset of that broader concept. That’s why famous systems built by OpenAI, Microsoft, or Google can interpret our conversations, make sense of them, read and summarize our documents, or even photos and images, and then respond in a way we humans can understand.

Natural language processing has developed at breakneck speed thanks to cloud computing and distributed processing systems. Likewise, access to huge volumes of data — millions of documents, photos, images, and videos — has enabled the buildup of a knowledge base substantial enough to let these systems “learn” far more.

Large Language Model — LLM

LLMs are the core of the artificial intelligence systems that have taken center stage in recent months. LLMs are models that, thanks to huge amounts of information — almost always text — are able to interpret, analyze, and rewrite or generate related information.

To understand how an LLM works, think of it like getting to know a person who has read many, many, many books, and yet isn’t particularly creative. This “hypothetical” person has a great ability to summarize, condense, and connect concepts from different books, but isn’t able to generate genuinely new ideas from the information they know.

Generative Artificial Intelligence

These previous concepts let us understand GenAI, or generative artificial intelligence: the kind capable of “generating” or “creating” content — text, images, and even video — from huge volumes of data inside an LLM.

Today the landscape has matured: ChatGPT is no longer the only reference point, and now competes with Claude, Gemini, and a growing family of “agentic coding” tools capable not just of generating code, but of running it, testing it, and fixing it on their own. The question is no longer “is there a GenAI tool that’s any good?” but “which one, for which task, and with how much oversight?”

Now that the concepts are clear, let’s talk about the future of work.

Automation and Augmentation of Work

There are two important concepts around using AI at work: automation and augmentation.

The first, automation, refers to tasks or duties that will no longer require any, or almost any, human interaction. Think of relatively simple tasks like summarizing a document, taking meeting notes, or analyzing a company’s financial data, managing inventory, or projecting its growth or decline. These tasks can be automated, and we humans will become “customers” of these services.

The second concept, augmentation, refers to human tasks that can be enhanced or have their reach and impact increased thanks to AI. For example, analyzing medical test results to help inform treatment or diagnosis decisions. While someone might think this could be 100% automated, it’s worth remembering that AI isn’t legally accountable for its decisions. That’s why the human factor can’t be fully removed — and I’m not sure we should try to.

Accountability

Before we move on to the future of work and which occupations are at risk, it’s worth talking about accountability. It doesn’t have a clean one-word equivalent in Spanish, but even in English, plenty of people confuse it with plain “responsibility.” I always like to use a simple example to mark the difference (if you’ve read my blog before, you’ve probably seen it).

A father walks into a store with his 8-year-old son. The kid starts running and shouting and, in a moment of carelessness, breaks something. The child, aware he misbehaved, apologizes to his father. The store owner charges the father.

In this short example, it’s clear the child is responsible, but the father is accountable. Responsibility and accountability carry legal and financial implications that many people still don’t anticipate.

If you’re still unsure about this, remember that just as a pet owner is legally and financially liable for any damage their pet causes, a company that uses AI-based systems is legally and financially accountable for the actions tied to it.

And we already have clear examples — just take a look at the Air Canada case, where the company had to legally and financially honor something its GenAI-based customer service system had promised.

Professions Exposed to AI

I think the meaning of AI and GenAI is a lot clearer now. So this is a good moment to look at the impact of AI on jobs and professions.

The World Economic Forum projects that by 2030, AI will create close to 170 million new jobs globally, while displacing about 92 million existing ones. The net is positive (+78 million), but that figure hides the most important part: the people who lose their jobs are almost never the same people who end up filling the new ones. Around 60% of occupations have partial exposure to automation — most jobs won’t disappear, they’ll get reorganized task by task.

Contrary to what most people might think, software development remains one of the most exposed professions — and today we have something we didn’t have in 2024: evidence of what’s actually happening inside that exposure. A Harvard study of 62 million workers found that when companies adopt generative AI, hiring of junior developers drops between 9% and 10% over six quarters, while senior developer hiring barely moves.

And that’s the problem that worries me most — as a teacher, even more than as a consultant: if we hire fewer juniors, and on top of that, the way they learn degrades, we break the chain through which senior developers have historically been trained. A 2026 Anthropic study showed that developers who use AI to delegate code (ask it to write it and move on) score up to 17 percentage points lower on comprehension tests than those who use it to probe concepts — asking why something works, not just asking for it to work. The biggest gap shows up, precisely, in debugging: the skill you need to spot when the AI got it wrong. And you can’t thoroughly review what you no longer thoroughly understand.

And professions that are more human, closer to and centered on human contact and relationships, are opening up more room in the future of work.

Human Contact

AI runs into natural resistance from humans, because we still enjoy human contact. The pandemic made it very clear that there are different paths to happiness, but that connection with others — real, deep connection with family or friends — is essential.

And without getting too deep into the debate, we can understand the rise in mental-health and mood-related conditions, and the movement of people toward places where life is “more traditional” and “gets back to basics,” as a strong expression of a need to connect, to belong, beyond the likes or the idealized view some social networks offer. We’re now seeing a different work dynamic tied to the end of 100% remote work and the “return to office.”

The Most Exposed

According to the research, the most exposed professions are tied to repetitive work, data analysis, and, in general, tasks that follow an easily replicable or improvable script.

The Least Exposed

At the other end of the spectrum, we have professions with little sensitivity to GenAI. Among them: counseling of all kinds, care-related work, physical performance (athletes or performers), and, in general, professions where human interaction — between providers and clients, professionals and families, or work or competitive teams — can’t (or we don’t want it to) be replaced by autonomous systems.

Imagine a soccer championship with no humans playing. Maybe that’s interesting from a scientific standpoint, but would it be appealing, or considered a real replacement for the soccer we have today? Does it make any sense to the species, or to fans of the sport?

Final Thoughts

Thanks for sticking around to read the whole article. Again, nothing too complex here, and surely with a few technical imprecisions, but written with the sole purpose of reflecting on our role as leaders, teachers, and members of a complex society — one that’s still full of challenges and hasn’t yet managed to resolve its differences, or transcend them.

We’re Still the Smartest — For Now

The human race is still the most intelligent species (system) around. With all our flaws, our supremacy isn’t threatened yet. Even the most advanced AI still needs us to learn, grow, understand, and form opinions.

That said, if the day comes when AI surpasses us — and I personally think it’s possible — it wouldn’t mean the end of life, or that we’d be under threat. We humans haven’t thought about getting rid of every cat or every dog just because they’re “less” than us. It’s been our fears, our least rational — least intelligent — thoughts that have led us to wars or to the extinction of other species.

If AI truly does surpass us, to me that would feel like having a great mentor, a great caretaker offering its superior intelligence to help us do better. A bit like the message behind that 2014 Wally Pfister film, Transcendence, where it’s our fears and our ignorance that always keep us from moving forward.

Avoiding Autopilot Mode

Another important reflection is about strengthening our critical thinking — avoiding “autopilot” mode, and assuming AI is already better than us and that everything it says is perfect.

I would never discourage anyone from using AI tools at work or in their personal life. The important question is whether the person using AI is just “passing the message along,” or actually adding value. That’s where the risk lies. Whose jobs are at risk of disappearing? Well, the first in line will surely be those who believe their job can be done by simply copying and pasting.

If you’ve ever ended up lost, or worse, with your car underwater because you trusted a navigation system, you know what I mean. We need to use our brains, no matter how tempting it is to leave “auto” mode switched on.

Intellectual Property and Usage Rights

We’re still a long way from settling this one. But let me pose a case. I (hypothetically) decide to train an LLM system using every novel by the late Nobel Prize-winning writer Gabriel García Márquez. Then I use that LLM model to write a new saga of books.

Who owns the rights? Is it actually legal? Could I legally make money off of usufructing someone else’s work?

And to make things more interesting: what if it wasn’t someone else’s books, but my own? Does a book that took no effort have any merit? Would we read books that no one wrote or imagined?

If you think we’re far from that, read it again.

Use It, Don’t Abuse It

The responsibility falls on all of us. Using an AI tool should EMPOWER us as a species, not “eliminate” us as users.