If I Had to Start Learning AI Today, I'd Begin With This

Nine months ago, at TOTEM we put together an introductory guide to artificial intelligence aimed at people who wanted to understand what all this is about without needing any technical background. We've shared it internally and with some clients, and over time we've realised that the foundations it covers are still relevant, even though the sector has evolved significantly since then.

That's why I wanted to share it here. If you're just getting started or you want to build a more solid base around what AI is and how you can begin integrating it into your daily life, it can be a useful starting point.

Note: The guide is currently only available in Spanish. If you'd like an English version, feel free to reach out through the contact form on this site or via the contact details in the guide itself, and we'll send it over.

Access the TOTEM guide

What I think still matters most

Beyond the specific content of the guide, there's one idea that I consider fundamental and that hasn't changed since I started learning about all this: artificial intelligence can make us better at what we already do well and can complement the areas where we're weaker, but for that to actually happen you need to use it with some intention rather than just in any old way.

The first risk I see, and I think it's a real one, is delegating everything to AI without thinking too carefully about the consequences. The convenience of always having a tool available that solves what you need in seconds can be very tempting, but if you systematically stop doing things on your own, you end up losing capability. What you don't practise gets rusty, and in a context where AI does everything for you, what gets rusty is precisely your own judgement to tell whether what it gives you is good, mediocre, or just plain wrong.

The second point, which I think is the one most often overlooked in conversations about this topic, is that the real way to get value from these tools isn't to always use them or always avoid them, but to alternate: there are tasks and projects that, when you do them on your own, turn out better than they would with AI's help, and there are others where the tool gives you an enormous boost that would take you much longer to achieve alone. Identifying which category each thing falls into is a process that takes time and requires experimenting, but it's the one that truly lets you get something out of all this in a way that adds to you rather than subtracts.

Why the foundations are still valid

It might seem contradictory to talk about a nine-month-old guide in a sector where models update every few weeks, but there are things that don't change as fast as the headlines suggest: what a language model is, how text generation works, how to write a good prompt, which types of tasks are most suitable for delegating to AI and which aren't. Those concepts are just as valid today as when we wrote them, and if you don't have them clear, all the layers of tools and updates that come on top become much harder to make sense of.

We're looking at a technology that could mark a before and after in many sectors, and hopefully for the better. But for that to happen on an individual level, I think you need to understand it from the inside first (even if only at a basic level) before you start relying on it for everything.

Once you have that foundation, the interesting part is watching how the technology keeps surprising you: I write about that in the evolution of AI never stops surprising me.

I hope the guide is useful. If you have any questions or comments, you can reach out through the contact details in the guide or directly through this site. Take care.