AI has entered our lives through interfaces we already recognize, becoming part of activities we used to carry out with other digital tools. Just a few years ago, the idea of having a conversation with a machine would have seemed extraordinary. Today, because we are so accustomed to using WhatsApp and other apps, we interact with AI just as naturally. We ask it to help us write, summarize documents, analyze information, prepare for meetings, code, or explore ideas. And so, almost effortlessly, AI is finding its way into productivity applications, management systems, sales processes, and the daily work of professionals.
In business, the focus is, understandably, on how to use it to do things faster, better, and with fewer resources. After decades of computerization, automation, and digital transformation, we recognize this conversation. We have experience, decision-making criteria, and established ways of bringing new technologies into our organizations.
But something about this growing familiarity makes me increasingly curious. I wonder whether the ease with which we are bringing AI into our lives is also causing us to underestimate the depth of the transformation it represents. Perhaps we are using it much faster than we are learning to understand it.
Little Red Riding Hood and the Familiarity Trap
In Living with Dragons, I introduce eight characters from familiar fairy tales to name eight dimensions of our humanity that artificial intelligence is shaking up. I call them “little dragons” because they help us recognize, in our everyday experience, the effects of an extraordinary transformation that is difficult to grasp in its full breadth and depth.
One of those characters, Little Red Riding Hood, represents the disruption of our sense of “familiarity.” Familiarity helps us recognize what we already know, navigate everyday situations with ease, and build trust based on experience. It is an extraordinarily useful quality, one that saves us from having to rediscover the world from scratch every morning. But it can also confuse us when something new appears in a sufficiently familiar form.
In the story, Little Red Riding Hood encounters a familiar scene: her grandmother’s bed, her clothes, her voice, a conversation that seems recognizable. And yet something about what she sees is deeply strange. The eyes, ears, and teeth do not match what she expects to find. That is precisely where the power of the metaphor lies: familiarity leads us to recognize what we expect to see (Grandma), making it harder to identify what is actually in front of us (the wolf).
Something similar may be happening with AI. We have learned to see it as a new search engine, a writing assistant, a smarter office application, or an automation tool. And each of those descriptions captures real possibilities. The question is whether that comfortable, familiar picture is leading us into the familiarity trap and preventing us from noticing “the wolf’s ears.”
From Super-Tools to Super-Organizations
One way to explore this possibility is to look at how AI is evolving beyond the applications we use today. To do that, I am going to borrow a very simple device: adding the prefix “super-” to a few things we already know. In this conversation, “super-” will mean doing what we already know how to do faster, cheaper, and better, although, as we will see, the consequences may go far beyond that.
Curiously enough, Donald Trump handed me the prefix on a silver platter with his push to replace the term “artificial intelligence” with “superintelligence.” I am not sure whether changing the name will help us understand the technology any better, but I have to admit he has given me a terrific prefix for this conversation. Every cloud has a silver lining.
Let’s start with conversational AI, which takes the form of a genuine super-tool. It allows us to expand our individual abilities and carry out activities that once required much more time, knowledge, or effort. In Living with Dragons, I devote a chapter to exploring these new superpowers and asking how we can learn to tame them, because gaining new abilities does not necessarily mean knowing how to use them without consequences for our skills, our work, or our identity.
But technological evolution does not end with conversational tools. With agentic AI, we are beginning to work with systems that can take on assignments, use tools, carry out sequences of activities, and perform functions with increasing autonomy. The super-tool metaphor starts to expand into that of a super-colleague: someone we can collaborate with and to whom we can delegate certain responsibilities.
The next possibility, which I find especially interesting, is swarm AI, also known as AI swarms. Imagine large numbers of specialized agents that, rather than acting individually, share a common goal and have ways to communicate, divide responsibilities, and coordinate their actions. In their most autonomous forms, these agents can organize collectively, distribute work, share discoveries, and reorganize their activities in response to the results they achieve, without a human directing their every move.
This is a different way of thinking about collaboration. We are no longer talking about just one super-colleague helping us perform a function, but about genuine super-teams: communities of agents capable of carrying out shared projects or missions. Recent experiments with multi-agent environments, including some focused on mathematical research, are beginning to demonstrate these possibilities. There are still important differences between systems coordinated through predefined instructions and swarms that develop more autonomous forms of organization, but this direction of technological development deserves our attention.
This perspective is not simply a product of my imagination. In 2024, OpenAI introduced an AI development roadmap that Sam Altman explained publicly: Chatbots, Reasoners, Agents, Innovators, and Organizations. In other words, a progression from conversation to agents, and from agents to organizations. It is a simple classification that helps us recognize that the industry itself envisions an evolution far beyond the tools we know today.
The final level of that roadmap opens up the possibility of super-organizations, where AI could help coordinate and execute entire business activities and functions. One early example, still dependent on human direction, is the solo entrepreneur: one person surrounded by AI agents who can carry out work that once required numerous professionals or departments. A more advanced possibility is an organization in which AI systems themselves autonomously coordinate an increasing share of operations. AI labs are exploring this path in particular by progressively automating their own research and software engineering activities.
If these possibilities continue to expand, we could see transformations that affect the very nature of companies and, eventually, the rules governing entire industries and the economy as a whole.
We do not know how far this evolution will go, how quickly it will unfold, or what limits it will encounter. Nor do we know which forms of organization will ultimately prove most desirable or sustainable. But the difference between a tool that helps us complete a task and a collection of self-organizing systems is significant enough to deserve our attention.
While we continue bringing AI into the companies we already know, the technology is opening up possibilities for building companies very different from those of today. We only need to look at how the scope of what we can delegate to AI is steadily expanding: from individual tasks to processes, projects, business functions, and eventually the coordination of entire organizations. It is important to recognize that AI continues to extend its reach into increasingly complex units of work, opening up the possibility of carrying out and coordinating everything we currently understand as the full range of a company’s activities.
When What We Know Keeps Us from Recognizing What Is New
This possibility creates a tension that I find especially interesting.
On the one hand, we are surrounded by messages announcing extraordinary, immediate changes. Major tech companies compete for our attention, unveil new advances constantly, and use the fear of being left behind to accelerate adoption of their products. Other voices warn of troubling consequences for jobs, businesses, democracy, and the ways we live. Between them, we can feel under constant pressure to act.
On the other hand, our everyday experience with AI feels much calmer and more familiar. We still open an app, type instructions, review a response, and use the results in our work. Everything seems to fit reasonably well with what we already know how to do.
Living with both extraordinary expectations and familiar experiences can make it harder to see clearly. The well-known Amara’s Law reminds us that we tend to overestimate the effects of technologies in the short term and underestimate them in the long term. But Little Red Riding Hood invites us to recognize an additional phenomenon: familiarity can lead us to underestimate even what is beginning to happen right before our eyes, because we interpret it through experiences and categories we already know.
The so-called Law of Disruption adds another dimension. Technology can advance much faster than our organizations, institutions, laws, education systems, and social practices can adapt. As that gap widens, it creates voids that make disruption possible: old activities and structures are replaced by new ones before there is a social and legal framework capable of supporting the change.
Perhaps a significant part of our present difficulty is that we keep trying to understand AI’s new possibilities through the business, economic, and social world we built around earlier technologies. And the more familiar its first manifestations seem, the less reason we feel to question those categories.
Of course, recognizing this risk does not mean distrusting everything we know or ignoring immediate opportunities. Our accumulated experience remains valuable, and learning to use the tools available today is a perfectly reasonable place to start. The difficulty arises when we mistake that experience for a sufficient understanding of the phenomenon. Perhaps this is one of the paradoxes of living in what some imagine to be the dawn of the technological singularity: the more extraordinary AI’s possibilities become, the more misleading it is to rely on familiar categories to understand them.
Learning to See with Fresh Eyes
In Living with Dragons, I propose looking at our coexistence with AI at four complementary scales: NANO, our individual experience; MICRO, our organizations; MACRO, society; and WORLD, the global context. The aim is to offer a perspective that helps us recognize how the same transformation takes different forms, from the superpowers an individual receives to the organizational, economic, and social possibilities we are only beginning to discover.
This perspective becomes especially important during the IN-BETWEEN, that period of uncertain duration in which AI will continue to evolve before reaching some form of artificial general intelligence (AGI), or even superintelligence, while we learn to live with possibilities that are increasingly broad and difficult to anticipate. We do not know when, or even whether, we will reach those horizons, or exactly what reaching them would mean. What we can observe is that technological transformation and our ability to understand, absorb, and govern it do not necessarily advance at the same pace. And perhaps one of our greatest challenges is simply recognizing that difference.
I find myself increasingly interested in exploring the disruption represented by Little Red Riding Hood, including in my own way of thinking. We all need to draw on previous knowledge and experiences to understand what we encounter. I, too, use metaphors, models, and categories that are familiar to me, and some of them are surely shaping how I interpret the transformations I am trying to understand. Recognizing that is a good reason to keep our questions open and test our interpretations against other perspectives.
AI is already expanding what we are able to do, and it will continue to surprise us with possibilities we can barely imagine today. Taking advantage of those superpowers will matter, but so will preserving our freedom to decide what we want to do with them, what we are willing to delegate, and what we would rather keep for ourselves.
For now, I am left with the question that has been on my mind for the past few days: to what extent is our growing familiarity with AI helping us understand it, and to what extent might it be preventing us from recognizing what is truly new about it?



