One of the main debates around the growing use of artificial intelligence is whether it will manage to take your job. The usual “romantic” answer is no, you are a human professional who carries the ability to make decisions based on experience and on factors that only a human can consider, but… is that actually true?
It is a slightly raw reality and a difficult one to accept, but as a marketer who has been 100% immersed in the development of artificial intelligence and the tech world for the last 3 years, I can assure you that there will be many jobs that AI will replace completely. Others will be amplified, and others will remain without a problem. Let me explain why.
This is no longer about ChatGPT or artificial intelligence in general. It is about the software we call agents, which use the smartest artificial intelligence models in the market and have the ability to do things, to take actions. Right now, you have probably already heard of ChatGPT or Claude, but if you have not heard of Claude Code, Cowork, or Codex (from ChatGPT), then you still do not know the full capability of these agents. ChatGPT or Claude on the web, the ones you are used to using, are VERY smart, but basically they only serve you to provide information, help you create a text here and there, or analyze a document. But what happens when you have ChatGPT installed on your computer, with access to all your files, emails, and any tool you need to give it? That is when you realize the full potential of the smartest models in the market.
I am going to tell you a reality you probably will not like. The smartest models from OpenAI and Anthropic are already smarter than you. Yes, you who are reading this. The main difference is in the context and memory that a human has, in the expertise and the ability to retain information. OpenAI and Anthropic are working on improving the memory systems these models can have, but the “raw” intelligence is already there. That is why those who know how to get the most out of these models and their tools know that everything comes down to the context and information you give to the artificial intelligence, to the LLM.
So what is going to happen to jobs? I think adoption is going slower than it seemed it would, AI has been advancing at an incredibly fast pace, and it has left most of Latin America a bit behind. I believe that little by little, medium and large companies in Mexico and LATAM will begin to adopt these tools, but they will not trust them until after one or two years of using them daily. That is when we will gradually see a growing percentage of companies replacing lawyers, accountants, administrators, salespeople, designers, and developers. All of these positions, replaced by AI agents.
Software development
Today, developers in the United States are starting to have problems. I think it has been very clear in the last 2 years, with the massive amounts of layoffs that top development companies have been making. Then there are startups, a world of small companies that recruit a small army of developers and sometimes grow into multinationals. This case is one of the very first that will take advantage of AI and replace or automate much of its work.
In Mexico and Latin America, demand for developers has been growing, both from Latin American companies and from American companies. Why? Because many startups prefer to first pay cheaper devs, with Latin American salary standards, before paying a dev in the USA. I have several Mexican friends who work for startups and have not seen the layoff problem. However, I think this sector is walking a tightrope, and it is a matter of 2 to 3 years before most of it can be replaced by artificial intelligence agents.
New opportunities; AI orchestrators.
If my vision is correct, I estimate that within 2 to 3 years we will see a new profile of positions in development companies and companies in general, very probably across all industries. That profile will be an “orchestrator” of artificial intelligence agents, something like an “IT” specialist focused on AI models, tools, and agents, whose job will be to execute, maintain, and review everything AI agents do inside a company.
This role is somewhat technical, with a way of thinking that resembles an engineer’s. The text produced by LLMs is the complete opposite of an exact science. You can ask the exact same thing to the same model repeatedly and its answer will very probably never be the same. That is why prompt engineering exists. It may sound a little peculiar, but knowing how to write so that an LLM responds or does exactly what you want is an art. That is why it is considered an engineering discipline: the task of knowing what to write and what data to give to an LLM to obtain the most accurate, intelligent response, or to make it behave exactly as we want.
How can I prepare?
I am not going to tell you to learn how to write code, or to learn how to develop LLMs. I am going to tell you to learn how the LLMs from the largest and most powerful companies in the world work: Anthropic, OpenAI, Google. Learn how to use their tools, how to write for their LLMs, and pay attention to what they are developing and what is coming next.
If you have a degree, combine your knowledge with the use of AI tools. Learn to use LLMs like a professional, not only by using ChatGPT, but by working directly with the raw model. How it behaves, what it is capable of, and what it is not capable of. By understanding this, you will understand where artificial intelligence development is going and you will be able to grasp how intelligent these models are.
In concrete terms: open X (Twitter), follow Sam Altman and Dario Amodei, follow the developers and designers from OpenAI, Anthropic, and Google Gemini. The rest will come to you naturally through the algorithm. Learn the basics of what LLMs are and how they work, read the public documentation and guides from OpenAI and Anthropic. Use the models, use them a lot, push them to the limit, experiment, and create.
This is an opinion article, written by Claudio González. Bachelor in Science of Business Administration and Licenciado en Administración de Mercadotecnia, current Brand Marketing Director at Sales-Hub. No artificial intelligence was used to draft, edit, or proofread the original Spanish article.