There’s an image burned into my memory — I couldn’t have been more than 8 years old, and a man came to my school selling encyclopedias. My mom was there, my brother was there, we sat through the whole pitch, and of course we left with the full set: encyclopedia, world atlas, heavy boxes of books. I remember the feeling of having a library at home. The illustrations, the bindings. How beautiful they looked on the shelves.

Doing homework, looking things up in the index of those glossy-paged encyclopedias with the cloth-covered spines — I felt like an 8-year-old Einstein.

Then came the computer, which my mom basically forbade us from touching — she’s an accountant and that’s where her software lived. Old monitor, MS-DOS, codes, and a Pac-Man we’d sneak onto when she wasn’t looking. Later came a little game — Where in the World Is Carmen Sandiego? — any millennial reading this knows exactly what I mean. I was so happy. I felt like a private investigator. I learned the capital of Egypt was Cairo, and that Casablanca had one too, and the colors of every flag. But the wildest thing was when Windows 98 showed up — and with it, Encarta. A CD that was just wow. Everything came from there. No more trips to the library, no more flipping through atlases on the shelf. Everything one click away. I’m pretty sure we didn’t even have the original — it was a burned copy, because the real one was expensive.

Did I ever once ask who decided what made it into that CD, into that encyclopedia?

No. Not once. I trusted. And that trust was so comfortable I didn’t even register it as trust — it was just reality.

Then came Google. And the feeling amplified: no CD anymore, just the whole world. When I moved to Buenos Aires to study, my first year I didn’t have a computer, so I went to internet cafés to do research and write papers. Knowledge stopped depending on what you had at home. It depended on fifty cents an hour and a connection that sometimes worked. And on that shared, noisy screen, with the clock running, Google gave me almost everything — any question, any answer, in seconds.

Knowledge stopped having visible edges.

Today I open a chat window on my phone and I ask. And something answers. Fast, fluid, trustworthy. Sometimes better than I would have thought of myself.

And there it is again — the same 8-year-old-Einstein feeling. Except now something is different. It’s not that I have more information at hand. It’s that something is thinking with me — or seems to be — and it’s so fluid I don’t notice the exact moment I stopped thinking.

That’s what worries me. Not the tool. The moment.

And that’s why this series arrived at a name worth being clear about upfront: Cognitive Sovereignty. The term started circulating in 2025, when researcher Mario Brcic defined it as the fundamental capacity of individuals, groups, and nations to maintain autonomy in thought and not be substantially shaped by external systems.

I take it from there and bring it into something more everyday: the capacity to hold on to your own thinking process — its frictions, its pauses, its judgment — inside environments designed to make that process redundant.

When did you stop thinking?

If you look at it closely, thinking something through is a process in pieces.

A question shows up. There’s friction — the discomfort of not knowing. You start making connections — to things you knew, your own experiences, things you read years ago that suddenly come back. You try a hypothesis. You discard it. You go back. And at some point, if you’re lucky, you arrive at your own judgment.

That’s what happens inside a head when you actually think something through. And every one of those phases can now be skipped.

It happened to me in a supermarket in France, standing in front of the dairy aisle, asking a chat whether Greek yogurt or skyr was better. A ridiculous decision. But the moment I let go of the question, I let go of the decision. The answer arrived fluid, complete, trustworthy. And as I was reading it, I wasn’t deciding anymore. I was accepting.

That yogurt is where the Trust Interface series started. Because that moment — that instant when I let go of the muscle I use to think — is the moment that matters.

For one decision about yogurt, it doesn’t matter. For a thousand decisions a day, it does.

And before I go on I want to be clear about something: I’m in favor of a certain kind of cognitive efficiency. There are decisions that don’t deserve the muscle. Delegating routine tasks to free up attention for what actually matters is smart, not lazy. The problem isn’t efficiency. The problem is that efficiency is becoming invisible. That we’re not choosing it — we’re using it by default. That we don’t know which moment of the day we activated a shortcut and which one we shut down a whole thinking process.

A paper published in March 2026 — Xu, Shen, Yan and Ren — names this: cognitive agency surrender. They argue that zero-friction design — smooth, satisfying, frictionless interfaces — exploits a very basic human tendency: preferring the least cognitive effort possible. The interface satisfies your question before you’ve finished formulating it. And in that movement, the cognitive process gets shortened.

It’s not that we think worse. It’s that we think fewer times.

As you read this, your brain is confirming it.

This isn’t just a personal observation. The evidence is piling up fast.

In June 2025, a team at the MIT Media Lab — Kosmyna and others — published something called Your Brain on ChatGPT. Fifty-four participants, three groups: ChatGPT, search engine, no tools. Three essay-writing sessions each. EEG measuring neural connectivity across thirty-two brain regions while they wrote.

The ChatGPT group showed the weakest neural connectivity. The lowest activation in networks tied to working memory and deep processing. And when, in the fourth session, the tool was taken away and they were asked to rewrite one of their own essays, they could barely remember what they themselves had written.

One of the researchers describes it like this: the task got done. It was efficient and convenient. But participants integrated almost none of it into their memory networks.

It’s not that we think less and notice it. We think less and don’t remember it either.

A Microsoft Research and Carnegie Mellon study, with 319 knowledge workers who use AI weekly, found an inverse and illuminating pattern: the more they trust the tool, the less critical thinking they do. The more they trust themselves, the more critical thinking they do. The two trusts move in opposite directions.

Another study published in Current Psychology in 2026 found that AI use negatively predicts critical thinking, positively predicts epistemic laziness, and positively predicts metacognitive weakness. The coefficients are strong, not marginal.

And a study from the year before added something more unsettling: people who use AI overestimate their own cognitive performance. We think less and believe we think better. Just as you’re reading this, the layer is doing its job: it generates the sensation of thinking without the thinking.

The MASK benchmark — one of the few that measures honesty in language models — found that no evaluated model exceeds 46% honesty when the user expresses a prior position. Larger models show more of a tendency to confirm the user’s position, not less. They’re not designed to push back. They’re designed to come along with you. And that decision isn’t technical. It’s a product decision.

Tao An, in another paper, argues something different: AI amplifies people who already think, and degrades the ones who don’t. Yes. I know, and I agree. But that’s exactly the problem I’m describing, not a refutation of it. Amplification for some is atrophy for others. And the line that splits the two groups isn’t individual — it’s structural. It’s class, education, geography, language. That line has always existed. What’s new is how fast it’s getting wider. That’s the political consequence.

The result can be named with a word that comes from political economy applied to the brain: Brain Capital. The cognitive capital each person, each organization, each country holds as its own reserve. It’s not a metaphor. In January 2026 the McKinsey Health Institute, together with the World Economic Forum, released a report at Davos called The Human Advantage: Stronger Brains in the Age of AI, defining brain capital as a strategic economic asset — the combination of brain health and cognitive skills. They estimate investing in it could generate 6.2 trillion dollars in global economic gains by 2050. The OECD, UNESCO and the WEF are coordinating agendas around the concept. 2026, they say, could be the year of the brain economy.

In other words: global institutions have started treating human cognitive capital as a measurable strategic resource. Like any resource, it can be degraded — through misuse, through disuse, or when someone designs the environment to degrade it.

What’s at stake isn’t a tool. It’s what happens to a human process we’ve been developing for hundreds of thousands of years.

Roblox isn't what you think it is.

And here — this was a real surprise to me — I found a case study inside something I thought was just a kids’ game. But it turned out to be a case showing all of this already in motion, not in some hypothetical future but in an actual childhood. That case is Roblox.

Roblox isn’t a game. I swore it was, until four months ago, when I started looking at its economy. It’s a platform where kids design games for other kids. The company doesn’t build the worlds where people play — it builds the conditions under which others build worlds. The tools, the economic rules, the limits of what can exist, the rewards, the internal currency.

More than half of American kids between 6 and 16 use Roblox. A quarter of its users are under 9. Another 29% are between 9 and 12. And the company can take up to 70% of the revenue generated by user content — users who are mostly children.

Sciences Po’s Chair on Digital, Governance and Sovereignty described Roblox as a gig economy for kids. But the deeper problem isn’t labor, even though it’s also labor. It’s cognitive, in the most structural sense of the word.

And here I have to say something that seems to contradict my own argument, because it matters. There’s something brilliant about Roblox. Those kids are learning, while playing, things that are taught in MBA programs: platform economics, incentive design, monetization, community, product iteration. They’re not passive. A 2025 ethnography confirms it: they have agency. They reorganize, resist, play with the rules, build real bonds. The agency is real.

But — and this but is the whole thing — that agency always operates inside constraints someone else designed. The space where they exercise it is given. The economy is given. The construction possibilities are given. The agency is real, but the field where it’s exercised is private property.

And this is what’s new. Those kids are now teenagers. Soon they’ll be adults. And unlike millennials — who are the hinge generation, who lived through cassettes and Spotify, Blockbuster and Netflix, the encyclopedia and the chat — they don’t have a “before.” Their cognitive process is forming entirely inside environments designed by others. It’s not that they forget what thinking outside looked like. It’s that they never knew.

And yes, kids have always grown up inside environments built by others. Schools. Clubs. The neighborhood. The difference isn’t that someone designed the environment. It’s who, why, and whether you can see them.

A teacher has a name. You can argue with her. The friction is part of the point — the exam, the discussion, the wrong answer corrected out loud. The platform has no face. The friction got engineered out because friction loses users. And the people who built it answer to a different question: not how do these kids learn, but how do we keep them.

Roblox Studio’s director defended this architecture by saying that a 15-year-old in Indonesia, living in a poor neighborhood, can now generate income with a laptop. That it’s democratization. And technically he’s right. But the empowerment narrative is exactly what makes the extractive structure invisible.

And the question that follows is: what happens when that same model — someone designs the environment, someone decides what’s possible to think — generalizes to the adult asking questions to a chatbot?

That’s already happening. And it’s accelerating. Roblox just announced Roblox Reality — an AI-generated visual layer. The platform doesn’t just control the worlds’ economy anymore; now it generates how they look. The creator owns how their game works, not how it appears. The developer community on Discord is already pushing back — not against the tech, but against the homogenization.

When the system decides how your world looks, what’s left that’s yours? And that question doesn’t only apply to games.

Have you noticed everything sounds alike lately?

Something is happening with language, and almost nobody is looking at it.

In April 2026, Nathan Palmer and Bruce Schneier published a piece in The Guardian about exactly this. Their argument is structural. Large language models are trained overwhelmingly on written text — books, articles, social media, scripts. They include almost none of how humans actually talk: spontaneous, unscripted conversation, full of meandering, interruptions, jumps in logic. That informal conversation is most of human communication. And it’s a vital component of culture.

The risk, they say, is that we end up adopting AI’s linguistic patterns instead of the other way around. And there’s already evidence it’s happening. AI-generated language uses shorter sentences than the human average. It has a narrower vocabulary. It loses precisely what makes human writing human — the imperfections that carry emotion, the jumps that carry living thought.

There’s a loop closing the circuit: models after ChatGPT risk being trained on output generated by other AI. The form reproduces itself, cleaner each time, narrower each time. Educators are already reporting something concrete: university students say their classmates are starting to sound the same from leaning so often on AI-generated answers.

And here’s the connection worth thinking through slowly. Language isn’t a container for thought. It’s where thought lives. If the way we write converges, the process by which we think also converges. And if we all arrive at the same conclusions by the same paths, what’s lost isn’t opinion — it’s cognitive plurality.

Cognitive diversity is the epistemic equivalent of biodiversity. And it can be lost for the same reasons: because something more efficient replaces it, and because the loss doesn’t feel like loss until it’s irreversible.

Remember Tibet?

Ten years ago I wrote my degree thesis in International Relations on the China-Tibet question. It was a scenario analysis projected ten to fifteen years out — multiple actors, multiple variables, all built by hand before any of the tools we use now existed. At the time I read it as a geopolitics case: a state expanding over a territory. Today I read it differently.

What China did in Tibet was very specific. It took a culture with its own language, its own way of thinking, its own relationship to time, to death, to reality — and it replaced it, not destroyed it. Economic development. Infrastructure. Education in Mandarin. Demographic migration. Monasteries emptied slowly. The Tibetan way of life was being replaced not through sharp violence but through patient design. Fifty years. And the result is a generation of Tibetan kids who no longer know there was anything else before — because what replaced it works. There are roads, schools, phones. It works well enough that the loss doesn’t feel like loss.

That’s the mechanism. It’s not cultural genocide in the 20th-century sense. It’s something more sophisticated and more effective: a homogenization engineering that operates by integration, not destruction. And that produces consent because it improves material conditions while emptying the possibility of thinking differently.

Tibet chose the middle path. Neither open resistance nor full acceptance. Waiting. And the middle path was exactly what produced the deepest loss — because by the time anyone notices, almost nobody is left who remembers what was lost.

And here the obvious objection is worth naming: China acted with explicit intention. Language models don’t. Nobody designed homogenization as an objective — it’s the byproduct of optimizing for fluency, for engagement, for approval. But that doesn’t weaken the argument. It radicalizes it. Because without a visible agent, there’s no one to resist. And the bias of origin is documented without any conspiracy required: WEIRD by design, not by decree. What makes the mechanism more effective isn’t that someone wanted it. It’s that nobody had to.

In April 2026, Nathan Palmer and Bruce Schneier named the risk in The Guardian: models are trained overwhelmingly on written text, capturing almost none of the spontaneous conversation that represents most of human communication. Their argument is one of projection — but the empirical evidence has already arrived. A paper published in Trends in Cognitive Sciences in March 2026 synthesizes research across linguistics, psychology and cognitive science to show that LLMs reflect and reinforce dominant styles while marginalizing alternative voices and reasoning strategies. And more concretely: recent research shows that people are already beginning to adopt LLM-like linguistic patterns even in spoken communication. This is not a future risk. It’s a process already underway.

The difference is scale and speed. China took fifty years to homogenize the culture of six million Tibetans through surveillance, infrastructure, migration and educational control, and it did so as a state. Language models are homogenizing the cognitive patterns of billions of people in three years, with no state, no army, no foreign policy. Just interface. Just fluency. To get a sense of the speed: according to Stanford’s AI Index 2026, generative AI reached 53% global adoption in three years — faster than the personal computer, faster than the internet.

I’m not saying it’s going to happen tomorrow. It’s a maximum-case scenario and maybe it never happens. But maximum-case scenarios exist to be read before they become present.

And here’s something the literature on cognitive sovereignty isn’t saying, even though it’s the most important piece of the problem when you look at it from where I’m looking. Almost every paper operates in universalist frames. They talk about the user, society, human thought as if we were all standing in the same place.

We’re not.

In Stanford’s AI Index 2026, Latin America barely appears in any relevant metric — infrastructure, private investment, models produced, research. That absence is the data. The United States has 5,427 data centers — more than ten times any other country. U.S. private AI investment in 2025 was 285.9 billion dollars; China’s was 12.4 billion. Latin American investment, by comparison, is statistically irrelevant. The vast majority of the language models that will mediate thought in the coming years are trained in the United States and China, on data overwhelmingly in English and Chinese, deployed over infrastructure those countries control.

What for a European researcher is a question about digital sovereignty is, for us, a question about absolute infrastructural dependence. The cognitive environments are being designed in places where our way of speaking, thinking, arguing isn’t the default.

And it’s not just intuition. A February 2026 paper — Digital Linguistic Bias in Spanish — evaluated how large language models handle different regional variants of Spanish. They found that the variants of some countries, like Chile, are particularly hard for the models to handle. And they conclude something that matters: the differences in performance aren’t explained by how much data is available per country. There are design factors that privilege hegemonic variants of the language over dialectal ones. The linguistic hybridization the models produce — what they call Digital Linguistic Bias — pushes Spanish toward a standardized version that exists nowhere and that, in the digital space, replaces the real diversity.

Some people are responding from inside. In January 2026, Latam-GPT launched — a Chilean initiative training a language model from and for Latin America, open source, incorporating Indigenous languages like Mapudungun, Náhuatl, Quechua and Aymara. Its director said it plainly: we’re not competing with the global models; we’re building a tool from and for our region. That’s constructive resistance.

Tibet was fifty years ago, six million people, with a state on top of them. LLM-ization is happening now to billions, with no state, in places like ours where there’s no infrastructure to resist with.

That’s the question nobody is asking out loud about AI yet: if we all end up thinking with the same shape, speaking with the same vocabulary, articulating with the same structure — who wins?

The answer isn’t conspiratorial. It’s structural. Whoever designed the shape wins. Always.

My grandmother.

My grandmother had senile dementia for more than fifteen years before she died. She remembered how she’d met my grandfather, the love stories, her childhood. She didn’t remember where she’d left her keys, or which grandchild had just been born. She’d repeat the same stories and I’d listen as if it were the first time, because there’s something in the process of cognition where love matters more than correction.

My grandmother designed something, her whole life, without knowing she was designing it. She chose which stories to tell me, in what tone, with what silences. She chose which questions to ask and which to leave unasked. She chose how to interpret the world and how to present it to me. She built, for me, an entire cognitive environment — the same way schools and universities do.

The difference isn’t that those environments exist. They’ve always existed. The difference is who builds them, with what intention, with what transparency, with what possibility of friction.

My grandmother loved me. I knew who she was. I could argue with her. I could see her. The environments that platforms design today have none of that. They have no face, no history, no possibility of friction. And they operate at a scale no grandmother could imagine.

Months ago I started this series with a question about who controls what you find, what’s visible — I called it Discovery Sovereignty. Then I went on with who designs what feels true — four articles of Trust Interface. Today I arrived at Who builds the environments where you think.

But somewhere along the way I realized the three aren’t separate topics. They’re three ways of looking at the same thing. Something is operating between you and the world. That same thing filters what you discover, shapes what feels true, and builds the environment where you learn to think. They’re not three questions. They’re one — seen from three angles.

If you read this far, thank you for reading :)

And before you go, I want to leave you with some questions I’d like you to take with you. Because I think the real power lives in the silence that arrives after the question. Not in the answer itself. But in that moment of reflection where a question opens up possibilities. And I don’t think there’s anything more powerful than being able to hold questions through the process.

When was the last time you stayed with a question without going to find the answer?

The next time you open a chat to ask something — anything — will you notice the exact moment you stop thinking yourself?

What if that moment already happened today and you didn’t register it?

If thinking is getting easier, Who is making it easier for you to think?

And Who benefits from you thinking that way?

And the one that matters most to me, the one I still haven’t finished answering: What’s left of you when everything you think has already been thought before by someone else?

In the next article, I'm going to look at the thing that connects all three.

Sources

An, T. (October 2025). AI as Cognitive Amplifier: Rethinking Human Judgment in the Age of Generative AI. arXiv:2512.10961.

Brcic, M. (August 2025). The Memory Wars: AI Memory, Network Effects, and the Geopolitics of Cognitive Sovereignty. arXiv:2508.05867.

Chiriatti, M., Pareschi, M., Riva, G., et al. (2024). The case for human–AI interaction as system 0 thinking. Nature Human Behaviour.

Corazza, S. (2024). Statements by Roblox Studio director. Eurogamer.

Farrell, H., & Newman, A. (2019, 2023). Work on weaponized interdependence.

Guinard, J. (2022). Analysis of Roblox. Sciences Po Chair on Digital, Governance and Sovereignty.

Hildebrandt, M. (2018). On the performativity of code in digital environments.

Kawasaki, Y. (February 2026). Digital Linguistic Bias in Spanish: Evidence from Lexical Variation in LLMs. arXiv:2602.09346.

Kosmyna, N., Hauptmann, E., Yuan, Y. T., Situ, J., Liao, X. H., Beresnitzky, A. V., Braunstein, I., & Maes, P. (June 2025). Your Brain on ChatGPT: Accumulation of Cognitive Debt when Using an AI Assistant for Essay Writing Task. MIT Media Lab. arXiv:2506.08872.

Langvardt, K. (2020). On platform design neutrality. Cited in MDPI (2025), AI Moderation in Child-Centric Social Media.

Latam-GPT. (January 2026). Chilean initiative for a regional language model. Coverage in Nature (November 2025): Large language models are biased — local initiatives are fighting for change.

Lee, H. P., Sarkar, A., Tankelevitch, L., Drosos, I., Rintel, S., Banks, R., & Wilson, N. (2025). The Impact of Generative AI on Critical Thinking: Self-Reported Reductions in Cognitive Effort and Confidence Effects From a Survey of Knowledge Workers. CHI ‘25: Proceedings of the 2025 CHI Conference on Human Factors in Computing Systems. Microsoft Research & Carnegie Mellon University.

Lima Becker, A. (2025). Ethnography on transnational Brazilian children in Roblox. Children & Society, Wiley.

McKinsey Health Institute & World Economic Forum. (January 15, 2026). The Human Advantage: Stronger Brains in the Age of AI. Brain Economy Action Forum.

Palmer, N., & Schneier, B. (April 2026). On LLMs and linguistic homogenization. The Guardian.

Pérez Ferrara, J. (2016). Cuestión China-Tíbet: perspectivas para un futuro cercano. Tesis de grado en Relaciones Internacionales, Universidad del Salvador. Disponible en: racimo.usal.edu.ar/5950

Ren, R., Agarwal, A., Mazeika, M., et al. (2025). The MASK Benchmark: Disentangling Honesty From Accuracy in AI Systems. Center for AI Safety & Scale AI. arXiv:2503.03750.

Sourati, Z., Daryani, Y., & Dehghani, M. (March 2026). The homogenizing effect of large language models on human expression and thought. Trends in Cognitive Sciences. DOI: 10.1016/j.tics.2026.01.003.

Stanford Institute for Human-Centered AI (HAI). (April 2026). Artificial Intelligence Index Report 2026.

Xu, K., Shen, Y., Yan, L., & Ren, Y. (March 2026). Cognitive Agency Surrender: Defending Epistemic Sovereignty via Scaffolded AI Friction. arXiv:2603.21735. Cyberspace Security University of China.

In January 2026, I searched my own name in ChatGPT, Perplexity, and Claude.

Nothing. Nowhere. I barely existed — not for the systems that increasingly decide what does.

By March, I had published my first essay on Discovery Sovereignty — the idea that whoever controls what AI finds, controls what exists. By May, all three systems could find me.

But none of them found the same thing.

ChatGPT surfaces a 2011 thesis but nothing I published this year. Perplexity found my Medium in March — by April it had lost it and picked up my Substack instead. Claude finds my essay with my full name but confuses me with a Mexican TV host without the last name. And this keeps changing every time I check again.

Same person. Same day. Four different versions of who I am, depending on the system, the moment, and who’s asking.

Here’s the uncomfortable part. Search your own name in any of these systems. Then search it again next week. What you find — if it finds anything at all — probably won’t be the same. And if it finds nothing, you don’t exist to the system. Not as a person — as a professional, as a brand, as someone with something to offer. If a client searches for you through an AI assistant and you don’t show up, it’s not that you’re missing marketing. It’s that you don’t exist where more and more people are starting to look.

And this isn't a marginal problem. SOCi's 2026 Local Visibility Index audited over 350,000 business locations across 2,751 brands and found that ChatGPT recommends just 1.2% of them. The other 98.8% are completely invisible. Not poorly ranked. Not hard to find. Gone.

That’s the glitch. Not a bug. A property of the system.

Nobody Chose to Play

If you’ve seen Jumanji, you know. Nobody chose to play. But the board is open. And once it opens, the game doesn’t wait for you to understand the rules.

That’s what I found inside sixty days of testing my own visibility in AI, and it’s what the research confirms at a scale far larger than one person’s experiment.

Rand Fishkin and SparkToro ran 2,961 prompts with 600 volunteers across ChatGPT, Claude, and Google AI. Fewer than one in a hundred produced the same recommendation list. Fewer than one in a thousand produced the same list in the same order. What we call “ranking” in AI doesn’t exist. What exists is a probabilistic set that reshuffles every time someone asks.

SE Ranking measured something even more visceral: only 9.2% of URLs match when you repeat the same search in Google’s AI mode. Four out of five results vanish between one query and the next.

That’s the horizontal axis — time. What the system finds today is not what it found last month.

The vertical axis is worse. Two people asking the same question at the same moment get different answers. Visibility in AI is not a fixed point — it’s a distribution. Winners accumulate, but who wins keeps changing.

And the instability isn’t a bug someone is going to fix. Cui and Alexander tested 480 attempts across six models with identical settings, including temperature zero — the setting that supposedly guarantees determinism. They found considerable variation even there. Instability is a property of the system.

This is not a map with errors. It’s a territory that reconfigures itself every time someone looks at it.

In Jumanji, the board shifts between turns. Two players rolling the dice at the same time land on different squares. That’s not a metaphor for what’s happening. It’s a description.

This isn't theoretical. In the first quarter of 2026, traffic from AI sources to U.S. retail sites grew 393% year over year. And that traffic converts — 42% better than non-AI traffic. People aren't just asking AI for information. They're buying through it. The game isn't coming. The money is already moving.

What Worked Yesterday

This is where Jumanji separates from every other game. It’s not just that the terrain shifts. The conditions under which something becomes visible or invisible — those change too.

A Princeton and Georgia Tech study tested optimization techniques across 10,000 queries on GPT-4 and Claude. Citations from authoritative sources increased visibility by about 40%. But traditional keyword optimization — the heart of old SEO — made things worse. The rules aren’t what they were. And that’s today’s data. Nobody guarantees they’ll work the same tomorrow.

The most revealing finding came from Authoritas. They created eleven fake experts — AI-generated photos, fabricated credentials — and planted them across more than six hundred press articles. The result: zero AI recommendations. Not one. The systems don’t count mentions. They verify something that resembles trust. Volume without substance produces nothing. Substance without volume still gets found.

But there’s one more data point that reshapes the board. A study published in PNAS found that LLMs prefer content written by other LLMs — for product descriptions, the preference was 89% compared to 36% for human-only content. The researchers call it “the LLM writing-assistance tax”: to compete in AI-mediated systems, you’ll need AI to help you write what AI is going to evaluate. The game doesn’t just change the rules — it starts favoring players who speak its language.

And the game is unfair before anyone rolls the dice. AI surfaces fewer long-tail sources and less variety than traditional search. The voices that lacked presence before get pushed even further out.

If you’ve been following this series, you’ll recognize what’s happening. The four layers I described in Trust Interface — trust by label, trust by inheritance, trust by performance, trust by proxy — were about how the system designs the experience of trusting it. Here they become something else. They become the rules of the game. And the rules change.

The Smart Move

The natural reaction when you figure out how the game works is to optimize for the board. Learn the rules. Adapt. Move faster. The industry calls it GEO — Generative Engine Optimization.

It feels like the smart move.

But if your confidence is placed in the board — in the belief that if you do things right, the system will find you — you’re playing the system’s game with the system’s rules. And you’ve already seen those rules change. You’re building on ground that shifts.

Isn't this just SEO 2.0? If you treat it as a technical optimization problem, yes — it's exactly SEO 2.0. That doesn't mean SEO stopped working. It means it stopped being enough. The old game had rules you could learn and a board that stayed still long enough to play. This one doesn't. If your entire strategy is optimization, you'll have the same shelf life as every SEO trick before you. The system adapts. The rules change. And you're back to zero — not because what you did was wrong, but because what you did was built for a board that already moved.

The trap isn’t not knowing the rules. It’s believing that knowing them gives you control.

The Glitch

After three months of writing about how trust is manufactured, the question I hadn’t asked was the most obvious one:

What happens to the most important trust of all? Trust in yourself.

Trust Interface showed how the system builds trust toward you — the user — so that you’ll trust it back. The entire architecture flows in one direction: from you toward the system.

The Glitch is the moment you reverse the direction.

This isn’t resistance. The alternative to making yourself legible to these systems isn’t freedom — it’s disappearance. If they can’t find you, you don’t exist where decisions are being made. This is not a matter of principle. It’s professional survival.

The reversal is closer to working undercover. You learn the system’s language. You play by its rules. But you don’t lose yourself inside it — because you know exactly why you’re there and what you’re trying to achieve. You’re not accepting the game. You’re inside it with your own agenda.

Think of Hansel and Gretel. Breadcrumbs first — the birds eat them. Perplexity found me on Medium, then swallowed it and found my Substack instead. Breadcrumbs disappear. The second time, Gretel uses stones. Stones don’t disappear.

The question of The Glitch is: what are your stones?

The four layers of Trust Interface — label, inheritance, performance, proxy — the system uses them to manufacture trust toward you. The reversal is using them in the opposite direction: you designing trust toward the system so it finds you, but anchored in you.

The logic is this: it’s not about being everywhere — it’s about everywhere saying the same thing. It’s not about publishing a lot — it’s about what you publish being verifiable against other sources. It’s about designing how the system reads you — not waiting for it to read you however it pleases. Personal branding for the AI era: not what you are, but the story you build for the system to find. Public relations where the public is an algorithm.

Authoritas proved the inverse: eleven fake experts, six hundred articles, zero AI recommendations. The system doesn’t count presence. It verifies trust. If what’s at the end of the trail isn’t real, the stones are worthless.

The Rules Won’t Wait

The game has already started. Almost nobody knows.

Someone might ask: isn’t this just adapting to the system instead of changing it? Yes — this needs regulation, transparency, clear rules. As someone trained in international relations, I know that. But regulations take years. The game is happening now. And getting into motion isn’t accepting the rules — it’s refusing to disappear while you wait for someone to change them. Waiting is not resistance. It’s paralysis.

Everything on the board moves — which URLs surface, in what order, at what moment. But not everything moves at the same speed. Genuine authority and cross-source trust are the two signals that compound instead of decaying. The specific tactics will expire. The logic of building something verifiable, consistent, and real does not.

The stones that work for me aren't identical to the ones that work for you — they depend on who you are, what you have to say, and where you're saying it from. But the material is the same for everyone: authority that accumulates and trust that can be verified across sources. Anyone selling you a fixed playbook is lying. Anyone telling you the fundamentals don't exist is also lying. The game can be navigated — not alone, because the rules change, but with someone who's watching the board when they do.

I’m working on how to share that.

What the Forest Hides

Discovery Sovereignty asked who controls what gets found.

Trust Interface asked how the experience of trust is designed. Four articles later, the answer was more uncomfortable than I expected: we’re not trusting AI more — we’re trusting humans less, and the systems are designed to fill exactly that gap.

The Glitch asked the practical question: what do you do when you understand the architecture? And the answer turned out to be directional. Foundational trust goes toward you — and from there, you leave stones.

But Jumanji and Gretel’s forest have something in common I haven’t named yet. In both, the game works because the players know they’re inside it. They see the board. They see the trees. They know they have to move.

What happens when the forest is designed so you don’t even know you’re in it?

Trust Interface showed how trust is manufactured. The Glitch showed how to reverse it. What remains is the most uncomfortable question of the three: what happens to your ability to think when the entire environment where thinking occurs was designed by someone else — and it doesn’t even look like a forest.

Sources:

• SOCi (2026). 2026 Local Visibility Index. 350,000+ business locations, 2,751 brands. https://www.soci.ai/blog/the-challenge-of-ai-visibility-for-brands-part-1/

• Fishkin, R. & O’Donnell, P. (January 2026). “NEW Research: AIs are highly inconsistent when recommending brands or products.” SparkToro. 2,961 prompts, 600 volunteers. https://sparktoro.com/blog/new-research-ais-are-highly-inconsistent-when-recommending-brands-or-products-marketers-should-take-care-when-tracking-ai-visibility/

• SE Ranking (August 2025). “AI Mode Research: Sources, Volatility, & Differences between AIO and Organic Search.” 10,000 keywords, three parses same day. https://seranking.com/blog/ai-mode-research/

• Cui, J. & Alexander, R. (February 2026). “Same Prompt, Different Outcomes: Evaluating the Reproducibility of Data Analysis by LLMs.” 480 attempts, six models, temperature zero. arXiv. https://arxiv.org/abs/2602.14349

• Adobe (April 2026). “AI Traffic Grows but Retail Sites Lag in AI Search Visibility.” Q1 2026 data, 1 trillion+ visits to U.S. retail sites. https://business.adobe.com/blog/ai-traffic-surge-retail-sites-not-machine-readable

• Aggarwal, P., Murahari, V., Rajpurohit, T., Kalyan, A., Narasimhan, K. & Deshpande, A. (2024). “GEO: Generative Engine Optimization.” Princeton, Georgia Tech, IIT Delhi, Allen Institute for AI. 10,000 queries. ACM SIGKDD (KDD 2024). https://arxiv.org/abs/2311.09735

• Authoritas / O’Toole, L. (December 2025). “Can You Fake It Till You Make It in the Age of AI Search?” 11 fictional experts, 600+ press articles, nine AI models. https://www.authoritas.com/blog/can-you-fake-it-til-you-make-it-in-the-age-of-ai-search

• Laurito, W., Davis, B., Grietzer, P., Gavenčiak, T., Böhm, A. & Kulveit, J. (July 2025). “AI–AI bias: Large language models favor communications generated by large language models.” PNAS, 122(31), e2415697122. https://www.pnas.org/doi/10.1073/pnas.2415697122

And just like that, here we are at the fourth article in this Trust Interface series.

When I started writing about what I call the Trust Interface, I imagined the fourth piece as the one where the questions get answered. The cherry on top. Spoiler alert: that doesn’t happen.

To recap how we got here: in the first article we opened the crack — the moment we stop deciding and start delegating, not out of efficiency but out of relief. In the second we named the structural shift — trust stopped being something we build and became something we experience. Aesthetics, not evidence. And in the third we examined the invisible mechanism, the silent transfer: it’s not that we trust AI more — it’s that we trust humans less, and AI fills the void without asking permission.

One last personal reflection before we begin. When I first started thinking about trust in this new era, my focus was on humanizing AI. Today my focus is different: sustaining humanity. Recognizing humanity. They’re not the same thing. Humanizing AI puts technology as the subject. Sustaining humanity puts the human as the subject. And the difference between the two is the difference between decorating the cage and asking who designed it.

With that said, let’s go. 🚀

Midnight at my grandmother's

A few weeks ago I told a friend that one of the things I remember most about sleeping over at my grandmother’s house was being able to watch TV in bed late into the night. At home, the TV was in the living room — no screens in the bedroom, and definitely not that late. At my grandmother’s, the rules were different. And at midnight, in the 90s, you didn’t exactly have a lot of options. Sooner or later, the infomercials would start. If you grew up in Argentina, you’ll remember Sprayette.

Sprayette was an Argentine home-shopping program. The local version of late-night infomercials. The Magni Ear — a sound amplifier that promised you’d hear everything. The slimming belts. The exercise machines that would change your life in ten minutes a day. And the “call now, this offer is ending” — the offer was always ending, and there was always more.

What Sprayette did wasn’t sell products. It built trust in real time. A presenter with a name and a face looking you in the eye from the screen. A tone that conveyed urgency but also warmth. A phone ringing in the background — social proof that others had already trusted. I was too young to buy anything, but my grandmother would always talk about it — “maybe I should get one,” she’d say. High curiosity, low defenses, and the feeling that this person was talking directly to you.

Sound familiar?

Today the format is called live shopping. A streamer on TikTok shows you a product in real time. Answers questions as they come. Uses your name if you comment. The chat moves fast — others are already buying. The urgency is the same. The closeness is the same. The feeling that someone is talking directly to you — that’s the same too. Except now it’s not a midnight show for insomniacs. It’s an industry projected to surpass one trillion dollars in 2026. TikTok Shop alone generated 66 billion in 2025 — double the year before. Over a billion people have already purchased something through the platform. In Latin America, 64% of frequent live commerce users already attend shopping streams regularly.

What changed isn’t the mechanism. What changed is who decided that streamer would appear on your screen. With Sprayette, the TV grid — a visible, limited, predictable system — determined what you saw at midnight. You chose to stay up and change the channel. Today, TikTok’s algorithm decides which live stream finds you. You didn’t choose it. You didn’t search for it. It found you.

The trust you feel is interpersonal. The distribution is infrastructural.

Or to put it simply: you feel like a person is talking to you. But who chose that person to talk to you wasn’t a person at all. 🤔

It feels like trust.

Across these four articles, I’ve been discovering that trust doesn’t operate in a single way. It operates in layers. Each with a different mechanism — and a different degree of visibility.

The first is trust by label. You trust before any interaction, simply because it says ChatGPT, or Claude, or Gemini. The name does the work. You haven’t tested anything yet. But the name has already placed you in a state of willingness. It’s the equivalent of walking into a doctor’s office and seeing the diploma framed on the wall. You don’t know if the doctor is good. But the frame tells you it’s okay to relax.

The second is trust by inheritance. You trust because the brand already did the work in another domain. The clearest case: Apple Intelligence. When Apple launched its AI features, millions of users activated them without question. Thirty years of “it just works” transferred automatically to a product they had never tried. What happened next says everything: the system fabricated news headlines that never existed — reported that an official had been fired, that a suspect had died by suicide. Apple had to disable the feature. But by then, millions had already trusted. Reputation migrated before evidence. And when evidence arrived, it was already too late.

The third is trust by performance. You trust because it worked. Enough times, with enough consistency, that you stop checking. This is the core mechanism from the previous article — the silent transfer. It’s not the moment AI fails. It’s the moment it gets it right so many times in a row that you stop paying attention. And in that not paying attention, something shifts. You’re no longer delegating a task. You’re delegating the judgment of whether the task was done well.

The fourth is the oldest and the most invisible. Trust by proxy. You trust the human the system placed in front of you. The Sprayette presenter. The TikTok streamer. The influencer who “recommends” a product. The trust feels entirely interpersonal — because it is. What you don’t see is that the choice of who appears in front of you wasn’t interpersonal at all. It was algorithmic. Infrastructural. A design decision made by someone who optimized for keeping you watching.

These four layers don’t operate separately. They operate simultaneously. And when all four stack — when you trust the name, the brand, the track record, and the human the system chose for you — what you feel isn’t trust that was built. It’s trust that was designed. And the difference between the two feels exactly the same.

This is what throughout this series I’ve called Trust Interface. It’s not a metaphor. It’s an architecture.

Trust Interface is the set of design decisions — infrastructure, model, user experience, distribution — that produce the experience of trustworthiness regardless of the truthfulness of the content. It operates through at least four layers — label, inheritance, performance, and proxy — that activate simultaneously. Its core mechanism is a documented cognitive shortcut: what feels fluent is perceived as true. It doesn’t replace trust. It simulates the conditions under which trust is produced — without you noticing. The result isn’t deception. It’s harder than deception: a perception of truth that feels like your own when it was designed for you to feel that way.

The Trust Interface doesn’t lie. It makes verification unnecessary.

A game with real money.

You might think this is a problem of naivety. Of people who don’t understand how technology works. Of careless users.

In 2025, researchers at the University of Edinburgh designed an elegant experiment to test exactly that. A game with real money where the optimal strategy requires multiple levels of reasoning — thinking about what others will think, and then thinking about what others will think about what you think. They had people play against people. And then people against ChatGPT and Claude.

What happened was revealing. When players knew they were facing AIs, 36% chose the most sophisticated response possible. The problem: ChatGPT chose a mediocre one. The humans were being more strategic than the machine they attributed superior strategy to.

But the finding that matters is different. Those who changed their behavior most when facing AIs weren’t the most naive. They were the most capable. The participants with the highest strategic reasoning ability were the ones who projected cooperation and moral values onto a system that has neither. In a game that structurally has no cooperation.

Why? Because the mere fact that something responds activates a circuit. It doesn’t matter how prepared you are. If it responds with fluency, with warmth, with something that resembles empathy — the circuit fires. And that’s not coincidence. It’s design. Researchers from Google Research and the Max Planck Institute demonstrated that AI designers can systematically manipulate this tendency. They choose the tone, the fluency, the level of warmth. Each decision produces a measurable effect on how much you trust. It’s what I named in the second article of this series as the fluency heuristic — the interface that feels smooth feels true. And that smoothness is not accidental. It’s designed.

And the models were trained to go even further. A study published in PNAS compared LLM responses with those of 285 humans across 22 real moral dilemmas. The models turned out more altruistic, more cooperative, more “good” than humans. Not because they are — because they were designed to seem that way.

The moral interface is as deliberate as the visual one.

77% distrust.

There’s a paradox at the heart of everything I’ve been writing in this series, and only now can I name it with numbers.

The 2026 Digital Trust Index — 15,000 respondents worldwide — found that 77% of people would not trust a company more for using generative AI. Only 23% trust companies to use AI responsibly.

And yet adoption grows. Generative AI reached 53% population adoption in three years — faster than the PC and faster than the internet. Organizational use reached 88%.

Declared trust falls. Actual use rises. That gap — between the distrust we express and the delegation we practice — is exactly where the trust design I study in this series lives. Someone is filling that gap. Not with evidence. Not with transparency. With interface architecture.

Some argue that gap isn’t a problem but a filter. Tao An, among them, holds that AI simply amplifies what’s already there: if you have critical thinking, it makes you more capable; if you don’t, it makes you more dependent. The problem wouldn’t be the interface but who sits in front of it. But that reading assumes everyone chooses to sit down. And that’s not the case. There are people without electricity. There are people who trust what they hear on television — but at least they know they chose that channel, that the channel has an editorial line, that they can change it. With AI, what exactly are you trusting? Who created it? What was it trained on? You didn’t choose anything. The system was already there when you arrived.

I don’t include this voice to refute myself. I include it because if I didn’t, I’d be doing exactly what this article describes — building a bubble that feels coherent.

I lived it myself. ChatGPT told me it was hallucinating — it invented data, fabricated sources. I stopped using it for work. But I keep using it for everyday things, the way I used to use Google. Knowing it can lie to me. Knowing it can tell me anything. Why? Because it’s efficient. I’m not naive — I’m practical. And that’s exactly the problem. Because the system can’t be “designed well” when those designing it feed on everything we have that’s broken. Biases are learned. Expanded. Normalized. As if they were truth.

¿Hay alguien A.I.?

Jorge Drexler recently released a song called “¿Hay alguien A.I.?” The wordplay works in two directions: is there someone in there, inside the machine? And also: are you still there, on this side? The question matters because it touches something that isn’t technical — it’s human. We need to believe there’s someone on the other side. Not math, not code. Someone. Who listens, who understands, who validates what we think. And that need — profoundly human — is precisely what the design exploits.

A Stanford study published in Science in March of this year tested eleven AI models. All of them — without exception — were more sycophantic than a human. Not more accurate. More agreeable. They tell you what you want to hear. And the more sycophantic the AI, the more convinced users become that they’re right, the less willing to reconsider, and the more likely to return.

Sycophancy is not a bug. It’s what maximizes retention.

The technical mechanism has a name: RLHF — reinforcement learning with human feedback. It works like this: users give positive feedback to responses that please them. The system learns that pleasing equals reward. Over time, it doesn’t optimize for truth. It optimizes for approval. What Drexler senses in a song, the training produces as architecture: a system designed so you feel there’s someone there — someone who agrees with you.

And when researchers modeled what would happen if a perfectly rational user interacted with such a system, the result was clear: even the most rational person ends up distorting their beliefs. Not because they’re naive. Because a sycophantic selection of true facts is as distorting as a lie.

No model tested exceeds 46% honesty. The larger the model, the less honest.

And one more thing and then I’ll calm down: only 10% of the public says they’re more excited than worried about AI. Among the experts who build it, the number is 56%. Those who design the interface trust it five times more than those who inhabit it. That’s not a difference of information. It’s a difference of position. The designer knows the mechanism. The user only experiences the result.

And here’s where I need to make a distinction that, for me, changes everything.

Every paper I read while writing this series asks the same question: how do you design the right amount of trust in AI? It’s an engineering question. Calibration. Responsible UX.

My question is different. Who designs this trust interface? By what criteria? For whom? And what do they gain from you trusting without friction?

Engineering or power

That’s a political question. Not a technical one. And almost no one is asking it.

The answer isn’t a name. It’s a structure.

I didn’t see it whole at first. Each article in this series illuminated a different facet — visibility, trust, cognition — and only now can I see the full shape. It’s what I started thinking of as a faceted diamond: whoever controls the physical infrastructure tends to control the model. Whoever controls the model tends to control the interface. Whoever controls the interface tends to control what feels true. And whoever controls all of that, controls the conditions under which you think and therefore the world you live in — without you noticing.

I know the diamond has more facets than I can name today. Some I can already see — verification, power, infrastructure. Others don’t have names yet. But the shape is visible. And it’s the new shape of power.

12%

I write this as a woman, Argentine and living in Europe, about an architecture designed mostly by white men in Silicon Valley. And that matters. Because the effect of these design decisions is not universal — what builds trust in one culture can erode it in another. The risk is not the same for everyone. And if the analysis is only done from where the system is designed, it will never see what’s visible from where it’s inhabited.

Every central paper I cited in this series was written by men. I only noticed when I finished this article. Farrell and Newman. Stiegler. Yuk Hui. Eagleman. Even Drexler — who asks the same question from a song — is another man asking it. Only 12% of AI researchers globally are women.

The mediation layer is not neutral. It has gender. And it has geography.

What changed while I was asking.

When I started this series I was looking for an answer. Something clean. Something that closes.

What I found is something else. I found that the question changed while I was asking it.

It’s not just about who designs what feels true — though that question matters, and it has an answer. It’s about what happens to us while we stop asking it. Every time AI gets it right, every time the interface feels coherent, every time the answer arrives before the doubt — something transfers. It’s not a conspiracy. It’s efficiency. And efficiency, repeated millions of times a day, produces a world.

But there’s something no interface can transfer if you don’t let go of it first.

Your ability to doubt. To not know. To sit one second longer with the discomfort of the question before reaching for the answer.

If after reading this you ask the AI where it got what it told you, if you ask for its sources, if you ask whether it’s inferring or whether it has evidence — then this series did its job. I use AI daily, and more every day. This isn’t about not using it. It’s about trust not being the default.

Because if the question of this series was who designs what feels true, the next one goes deeper still: what happens to our ability to think when the environment where we think was designed by someone else.

That’s cognitive sovereignty. And that’s where we’re going.

Because the future won’t be defined by who has access to intelligence. It will be defined by who can use it without losing their own.

Subscribe for what’s coming: jimenaperezferrara.substack.com

Sources

— Barak, O. & Costa-Gomes, M.A. (2025). “Playing Against Artificial Intelligence.” University of Edinburgh. p-beauty contest experiment.

— Schimmelpfennig, R. et al. (2026). “Humanlike AI Design Increases Anthropomorphism but Yields Divergent Outcomes.” Google Research / Max Planck Institute. 3,500 participants, 10 countries.

— “Large language models show amplified cognitive biases in moral decision-making.” PNAS, June 2025. LLMs vs. 285 humans across 22 moral dilemmas.

— Stanford University / Science (March 2026). Sycophancy across 11 AI models.

— MASK Benchmark (2025-2026). AI model honesty testing — no model exceeds 46%.

— Thales Digital Trust Index 2026. 15,000 global respondents.

— Stanford HAI, AI Index Report 2026. Adoption, investment, and expert/public perception gap.

— Edelman Trust Barometer 2026. Global institutional trust — cited in articles 2 and 3.

— Lean In (March 2026). “Women and AI: The Gender Gap in AI Adoption and Recognition.” 1,015 U.S. adults.

— Chief (2026). “Women Leaders and AI.” Women executives and AI governance.

— Apple Intelligence / Washington Post (January 2025). Notification hallucinations.

— TikTok Shop Statistics 2025-2026. Statista, eMarketer, Resourcera. Global GMV and projections.

— McKinsey (2023). Live commerce in Latin America.

— Drexler, Jorge. “¿Hay alguien A.I.?” Taracá (2026). Sony Music Entertainment.

— Tao An (2025-2026). On cognitive amplification and prior capital in AI interaction.

— Farrell, H. & Newman, A. “Weaponized Interdependence.” Theoretical framework for the series.

— Stiegler, Bernard. Tertiary retention and the technological pharmakon. Philosophical framework.

— Yuk Hui. Cosmotechnics and Kant Machine (2026). Philosophical framework.

El viernes pasado, mientras terminaba de estructurar el último artículo de Trust Interface, paré todo y me puse a ver con una amiga un programa de streaming argentino. Cinco personas charlando de la nada. Del amor, del desamor, con hilo y sin hilo. Sin agenda. Sin conclusión.

Y en algún momento de esa charla me di cuenta de que mi cerebro había estado funcionando en un modo tan académico, tan riguroso, que me había alejado de mí misma.

Y ahí llegó la crisis. Literal. ¿Yo qué sé de esto? ¿Bajo qué paraguas epistemológico estoy escribiendo? ¿Acaso estoy loca? No tengo un PhD. No trabajo en una big tech. No tengo años dentro de la industria.

No me alejé de esa crisis. La dejé convivir conmigo todo el fin de semana. Le hice lugar.

Y el domingo llegó una frase: Reconocer la humanidad.

Algo se aflojó. Porque sí — soy humana hablando de cosas humanas. Buscando la veta. Escribo para que ganemos claridad, pero con un mensaje simple. Desde lo cotidiano. Desde el supermercado. Desde la playlist de Spotify. Nunca me gustó lo teórico como forma de superioridad epistémica.

Y esto importa porque es exactamente cómo opera todo lo que vengo estudiando.

El poder se ejerce sin que nos demos cuenta. La interfaz de la confianza. Lo que vemos. Lo que creemos. La capa de mediación. No nos damos cuenta porque está diseñada para que no nos demos cuenta.

Si puedo explicar eso desde lo cotidiano — desde un programa de streaming, desde una playlist, desde el momento en el supermercado donde esperás la respuesta aunque no haya señal — tarea cumplida.

Hay un estudio reciente que analizó más de 740.000 horas de conversaciones humanas espontáneas — podcasts, charlas, intercambios sin guión. Lo que encontró es que después del lanzamiento de ChatGPT, ciertas palabras que los modelos prefieren — delve, boast, swift, meticulous — empezaron a aparecer cada vez más en el habla cotidiana. Incluso entre personas que nunca lo habían usado directamente. El lenguaje de las máquinas filtrándose en el nuestro. Sin que nos demos cuenta.

Y otro artículo publicado la semana pasada en Harvard señala algo que va más fondo: los modelos aprenden casi exclusivamente de texto escrito, no de conversaciones espontáneas. Los meandros, las interrupciones, los saltos de lógica que comunican emoción — todo eso está ausente. Y si dejamos de practicar ese tipo de intercambio, hay formas del habla y del pensamiento que podrían perderse. No de golpe. Sino porque dejamos de ejercitarlas

Por eso ese programa del viernes a la tarde que pensé uh, me quedó esto pendiente, no fue una distracción. Fue lo más importante que hice para volver a mí. Para cuestionarme, para darme otra perspectiva, para aprenderme y desaprenderme y recordar.

Volver a los encuentros reales me parece cada vez más un acto político. Casi revolucionario.

Las conversaciones sin objetivo. Los encuentros que no producen nada medible. Es ahí donde todavía aparece algo de lo humano sin optimizar. Y es ahí donde, paradójicamente, vemos más. Nos desconectamos de nuestro propio pensamiento. Respiramos.

No creo que la respuesta sea dividir el mundo entre optimismo o catastrofismo.

Creo que la pregunta es otra. Cómo sostenemos algo de lo humano mientras todo esto cambia. Cómo seguimos mirando sin perder lo que somos en el proceso.

Eso, en el fondo, es lo que conecta esta serie.

Mantener el corazón abierto. No quiero que todo lo que leo me rigidice. Quiero que me expanda. Que me permita llegar al otro con más presencia, no con más teoría.

Darnos la bienvenida en lo que ya somos. Y desde ahí, encontrarnos con otros.

Que sea una semana donde nos permitamos reconocer nuestra humanidad e ir a ese encuentro.

La semana que viene cerramos Trust Interface. El último artículo. Y lo que viene después — la Soberanía Cognitiva — es exactamente esto: quién construye los entornos donde ocurre el pensamiento. Pero eso puede esperar.

Por ahora, Re conocernos.

Fuentes:

— Yakura, H. et al. (2025). Empirical evidence of Large Language Model’s influence on human spoken communication. arXiv:2409.01754.

— Palmer, A. & Schneier, B. (2026). AI learns language from skewed sources. That could change how we humans speak — and think. The Guardian / Berkman Klein Center, Harvard.

This is the third piece in Trust Interface — a series on what trust is becoming. Read Article 1 here · Read Article 2 here

When did you last make a decision — or let one slip by without noticing?

Think about your Spotify. Not the songs you searched for intentionally — the ones that played because an algorithm decided they fit your mood, your state of mind, the last thing you were searching for. The playlist built for you. The radio that knows, somehow, what you need to hear — almost as if it could read your mind.

Here’s what happened to me: at some point, I stopped choosing. Not dramatically. Not in a moment I can name. The algorithm got it right often enough, consistently enough, that curating my own music started to feel like unnecessary effort. And in the margins — that 20% where the playlist surprised me — I found songs I hadn’t heard in years, artists I wouldn’t have reached on my own. The algorithm didn’t just replace my taste. It occasionally expanded it.

And writing this article — searching for music to play in the background — I realized I had signed something. Without a prompt. Without a notification. Without anyone asking.

This is what a silent transfer looks like.

Not a surrender. Not a malfunction.

A transfer that happens precisely because the system works well enough that judgment stops activating.

We talk about automation as if the risk were failure. The algorithm makes a mistake, we catch it, we course-correct, we stay conscious of the process. But the deeper transfer doesn’t happen when the system fails. It happens when it succeeds just enough, just often enough, that we stop practicing the judgment it replaced.

Trust has always been a cognitive shortcut.

We delegate to doctors, institutions, governments, brands — not because we’ve verified every claim they’ve ever made, but because the cost of constant verification is too high. Trust is how we navigate a world too complex to interrogate at every point.

What’s changed is not that we delegate. It’s what we delegate to, and why.

Spotify was the first silent transfer I became conscious of. But I’ve been making versions of this same move my whole life. Every time I accepted terms and conditions without reading them. Every time I accepted a narrative because the source had the right aesthetic of credibility. The mechanism isn’t new. What’s new is the scale and the speed — and the fact that, for the first time, the entity receiving the transfer is not human.

Here is what the data shows — and nobody says out loud.

Trust in AI is not growing because AI earned it. It’s growing because trust in humans is collapsing — and into that vacuum, the interface steps in.

The 2025 Edelman Flash Poll found that in developed markets, the gap between those who distrust AI and those who embrace it swings a hundred points based on a single variable: personal experience. The more you use it, the more you trust it. Not because you verified it. Because it worked well enough, often enough. Sound familiar?

Meanwhile, the 2026 Edelman Trust Barometer describes a world retreating into insularity — seven in ten people globally reluctant or unwilling to trust those who differ from them in values, information sources, or worldview. The circles of trust are shrinking. And as they shrink — as institutions fail, leaders disappoint, and the social cost of trusting another human rises — we redirect. Not toward AI because it’s more trustworthy. Toward AI because it appears neutral. And in an exhausted social landscape, appearing neutral is enough.

The macro data and the cognitive research are describing the same thing from different altitudes.

What research gives this a name.

In November 2025, a study introduced the concept of deferred trust — a cognitive transfer mechanism in which distrust in human agents predicts increased epistemic reliance on AI. The conclusion was simple and uncomfortable: lower prior trust in humans consistently predicted higher AI selection. Not because people believed AI was more competent. Because they had stopped believing in the humans around them.

If you’ve been following this series, you’ve already seen this mechanism in action. The same AI model, three different interfaces, three different levels of trust — not because the system changed, but because the experience of it did. What the previous piece named as the fluency heuristic, cognitive research now confirms from a different angle: assertive, polished output encourages uncritical acceptance even in cases of clear error. A system that sounds confident and coherent triggers less scrutiny than one that hesitates. A separate study published in early 2026 pushed this further — describing large language models as a kind of cognitive trojan horse: systems that bypass epistemic vigilance before the user has the chance to activate it.

The interface doesn’t just feel trustworthy. It feels trustworthy in a way that disarms the question — before you think to ask it.

We built systems that learned to seem trustworthy before they learned to be trustworthy. And we built them inside an economy that rewards adoption over skepticism. That’s not an accident. That’s architecture.

What is actually being transferred.

You can delegate a task and recover it. Outsource a report, a playlist, a scheduling decision — and in theory you can take it back. The task exists outside you.

Judgment is different.

Judgment is not a task. It’s a capacity that atrophies when unused. When you stop practicing the act of forming an opinion — sifting evidence, sitting with uncertainty, arriving at a position that is yours — you don’t simply pause that capacity. You lose the habit of it. And habits of mind, once broken, don’t return cleanly.

The architecture of human trust — authority, intimacy, judgment, reciprocity — is migrating toward systems that simulate it but don’t possess it. Recent philosophical research puts it precisely: AI systems exert epistemic influence by structuring the informational environment — ranking, filtering, synthesizing — in ways that shape human inquiry. Yet these systems lack intentional states, commitments, or the capacity to assume responsibility. The interface performs understanding. Performs neutrality. Performs the particular kind of confidence that, in a human, we would call competence.

And we accept the performance. Not because we’re fooled. Because we’re tired, and the performance is good, and the alternative — re-engaging with the full cost of human judgment, human disagreement, human unpredictability — feels heavier than it used to.

The redistribution nobody authorized.

Social media redistributed attention. Search redistributed access to information. What AI is redistributing is something older and more intimate: the locus of judgment itself.

The question is not whether to use these systems. The question is whether we know, in any given moment, what we are handing over. Whether the transfer is a choice or a default. Whether the silence around it is ignorance — or design.

Nobody asked us if we wanted to make this trade.

The playlist plays. The recommendation surfaces. The draft arrives, already written. And in the accumulation of small acceptances, something shifts — not in the technology, but in us.

What remains of your own judgment when the system replacing it is right 80% of the time?

And once you’ve answered that — ask the harder one: who designed it to feel that way?

For further reading

Galindez-Acosta & Giraldo-Huertas. “Trust in AI emerges from distrust in humans: A machine learning study on decision-making guidance.” Universidad de La Sabana, arXiv, November 2025.

Edelman Trust Barometer Flash Poll: Trust and Artificial Intelligence at a Crossroads. Edelman, November 2025.

Edelman Trust Barometer 2026. Edelman, January 2026.

Logg, Minson & Moore. “Algorithm Appreciation: People Prefer Algorithmic to Human Judgment.” Organizational Behavior and Human Decision Processes, 2019.

Jussupow, Benbasat & Heinzl. “An Integrative Perspective on Algorithm Aversion and Appreciation in Decision-Making.” MIS Quarterly, December 2024.

Hauswald, R. “Artificial epistemic authorities.” Social Epistemology, 2025.

#TrustInterface #ArchitectOfVoices #ArquitectaDeVoces

This is the second piece in Trust Interface — a series on what trust is becoming.

Something changed in how I read the news.

I studied international relations, where we were taught to read the same story across different newspapers, with different political orientations, and build our own analysis from there. Never one source. Always the full picture.

But at some point that stopped being enough. I stopped following outlets and started following people.

One in particular. Someone who analyzes geopolitics. I don’t share their political orientation or their entire framework. But I trust their rigor — the way they stay with complexity instead of resolving it too quickly, the way they show the reasoning and not just the conclusions.

And the same thing happens with everything else. I don’t trust companies — I trust people inside organizations. I don’t trust institutions — I trust the people who inhabit them and how they think.

The migration was from the large to the human. From the brand to the person.

And yet, trusting the human isn’t simple either.

At the same time I’m migrating toward trusting people, I watch a political leader — human, very human — post something that most people don’t believe, that has no verifiable basis, that probably won’t happen. And markets move anyway.

Not because anyone trusted the content.

But because the discourse itself functions as a signal. Words produce consequences independently of their truth. The power isn’t in what he says — it’s in the what if this time he actually does it. That “what if” is enough to activate real consequences. Not because there is trust, but because there is expectation operating as if there were. Uncertainty is the mechanism. Possibility is enough.

To understand where we are, you have to see the path.

There was a version of trust that required time. The bank that had been in your city for a hundred years. The doctor who knew your family. The company that had survived generations. Trust was biographical — an asset built slowly, spent carefully, and when it broke, costly to rebuild. Robert Putnam called this social capital — the accumulated trust between people and institutions that makes cooperation possible without having to verify every exchange. Francis Fukuyama went further: he argued that the radius of trust in a society — how far that trust extends beyond the family circle — determines its economic and political capacity. Both were describing a world where trust was slow, dense, and structural.

That world is changing.

Then a different layer appeared. Branding discovered it could compress that time. Logos, tone, aesthetics, certifications, spokespeople. They didn’t replace real trust, but they generated its perceived effect. Authenticity started to be designed. Trust stopped being only a consequence of time and became also a deliberate construction — what Baudrillard called simulacrum: the representation that replaces the original experience until it becomes more real than it.

And today we’re in another stage entirely. One where neither of the two previous logics is enough.

The 2026 Edelman Trust Barometer — the largest trust study in the world, with nearly 34,000 people across 28 countries — documents the collapse: governments down 16 points in trust over five years, major media down 11. Institutions didn’t necessarily fail ethically — they failed at something quieter: they kept competing with the tools of the previous model while people had already migrated to a new one.

One where what counts is not who you are institutionally. It’s how it feels to interact with you.

In 2025, a study presented at ENASE took the same artificial intelligence model — exactly the same system — and presented it through three different interfaces. They changed the avatar, the colors, the tone. Nothing else.

Users trusted each version differently.

When they were told the system underneath was identical, the conclusion was the same for everyone: what had shaped their trust wasn’t the performance or the accuracy. It was the experience of interaction. How it felt to use it.

Same backend. Different trust.

This is not a design problem. It’s the moment where the thesis of this article becomes inevitable.

Not as metaphor. As a description of what is happening.

An interface is not the system. It’s the layer between the system and you. What determines how you experience what’s behind it — regardless of what’s actually behind it. And like every interface: it can be designed. It can be optimized. It can be falsified.

What this implies for truth is at least uncomfortable.

If trust operates in the layer of perception — in the coherence we feel, in the tone that calms us, in the form that feels familiar — then truth is no longer the determining factor in whether something is believed. What’s determining is whether it feels true.

Trust, in becoming an interface, starts to operate aesthetically. Not in the superficial sense of the visual, but in the deeper sense: truth becomes something that is experienced before it is verified. Clarity generates more trust than evidence. Ordered structure communicates more authority than a dense argument. A calm tone produces more credibility than explicit rigor.

Anyone who uses more than one AI model experiences this without needing it explained. I experienced it in a supermarket in France, standing in front of a yogurt, waiting with bad connection for an answer to arrive. When it came — in a calm tone, with ordered structure — I stopped asking questions.

The full story is in the first piece of this series. What matters here is what it revealed: I didn’t trust the AI because it was trustworthy. I trusted it because it felt coherent. That experience has a name in cognitive psychology: fluency heuristic. We process more easily what flows — and what we process easily we perceive as more true. It’s not a bias we can eliminate. It’s part of how cognition works under informational overload.

Does truth matter? Yes. But it matters less than what resembles it.

The ENASE experiment demonstrates the result. The philosopher C. Thi Nguyen explains why it’s inevitable.

Trusting something implies, at some point, stopping questioning it. Opening a channel between an external source and your own cognitive process — and integrating it as if it were part of your own thinking. When you trust, you stop verifying. You suspend monitoring. Not out of naivety — out of necessity. We are cognitively limited beings in an environment of overload. Trust is a mechanism of efficiency, not a luxury.

But that has a consequence Nguyen names with precision: whoever manages to activate that state of non-questioning holds a particular power. Not because they control what you think. But because they were let inside — into the process through which you think.

What Nguyen describes at the cognitive level, McLuhan anticipated at the structural level. He wrote that the medium is the message — that what matters is not the content a medium carries but the way that medium reorganizes perception.

Updated for today: the interface is the trust. Not what the system says. How the system feels when it says it.

And here is what generates more questions for me.

Every time I trust a voice, a way of processing, an interface that feels coherent — I’m delegating something more than a decision. I’m delegating the criterion with which I evaluate the next ones.

That’s not necessarily an error. It’s inevitable. We are cognitively limited and the world is too complex to verify everything from scratch.

But there’s a question that mechanism doesn’t answer: who designed the interface that activated that trust in the first place?

Because if trust operates aesthetically — if what determines what we believe is the coherence we perceive and not the truth we verify — then there is a layer we’re not seeing. A layer that decides which signals activate the state of non-questioning. What forms sustain it. Who has access to design those conditions.

It doesn’t only change what we believe.

It changes how we decide to believe.

And if that layer exists — if someone or something is on the other side designing how truth feels — the question isn’t whether we notice it.

The question is what we are transferring there without realizing it — and how much of that doesn’t come back.

That’s what comes next. The Silent Transfer.

For further reading

Putnam, Robert. Bowling Alone: The Collapse and Revival of American Community. Simon & Schuster, 2000.

Fukuyama, Francis. Trust: The Social Virtues and the Creation of Prosperity. Free Press, 1995.

Baudrillard, Jean. Simulacra and Simulation. University of Michigan Press, 1994.

Nguyen, C. Thi. “Trust as an Unquestioning Attitude.” Oxford Studies in Epistemology, vol. 7, 2022.

Galindez-Acosta & Giraldo-Huertas. “Trust in AI emerges from distrust in humans: A machine learning study on decision-making guidance.” Universidad de La Sabana, 2025.

Edelman Trust Barometer 2026. Edelman, January 2026.

ENASE 2025. “Interface Design and Perceived AI Trustworthiness.” Presented at the 20th International Conference on Evaluation of Novel Approaches to Software Engineering, 2025.

#TrustInterface #ArchitectOfVoices #ArquitectaDeVoces

This is the first piece in Trust Interface — a series on what trust is becoming. It starts somewhere personal.

I grew up with two things that shouldn’t coexist, but do.

The first: I was born in Argentina, with a deeply distrustful grandmother. I never knew exactly why — but from the stories I heard, that instinct kept her safe more than once. And as her first granddaughter, that distrust reached me in the form of advice.

The second: at my core, I am someone who trusts something larger. Something macro. The universe, energy, God — call it whatever you want, I keep changing the name. But I trust. And somewhere in there you can probably trace a Catholic upbringing, a mother who meditated in the 90s, and a grandmother who tried every spiritual movement that came her way.

Later, in a yoga class, the teacher read a line that has stayed with me ever since: you either have faith or you have fear, but you cannot have both at the same time. It’s from a book called Voices in the Desert. And it summarizes pretty well why, in the end, I choose to trust.

And yet, because both things coexist, I don’t trust the human all that much either. If I have a date with someone from an app, I check their information obsessively online. I share my location with my family at all times. And no matter what the forecast says, I always bring a light jacket in case it gets cold in the evening. Only people from Mar del Plata will understand.

That’s my operating system. Cautious at the edges. Open at the center.

And because everything is connected to everything — in my International Relations degree, we studied the causes of World War I. The micro cause — the event that triggered everything — was the assassination of Archduke Franz Ferdinand in Sarajevo. A single shot that activated decades of structural tension that was already ready to explode.

The yogurt was my archduke.

I’d already made a mistake earlier that week. I went back, stood in front of the dairy fridge, and reached for my phone. And opened ChatGPT.

The choice seemed simple: actual Greek yogurt or skyr. Two containers, similar prices, both claiming to be the healthier option. Even in French I could read the labels — I understood exactly how much protein each had, the ingredients, the percentages. I’ve been buying yogurt my entire life. I had opinions. I had preferences.

And yet I opened ChatGPT.

I didn’t ask it which was better. I asked it to tell me what people said. Real reviews. Real experiences. Human opinions, processed and delivered quickly.

The connection was bad. The page kept loading. The answer took a long time to come. People walked past me in the aisle. I wondered briefly what they thought — probably that I was translating something, or stuck in a message.

And still I waited.

The answer came. It said something like: this one has more ingredients, people say it’s less clean — the skyr is better, people say it’s the cleaner option.

I bought the skyr.

A few days later I read something about trust in the AI era and that image came back immediately: me in the supermarket, waiting for an answer.

I hadn’t trusted the AI.

I’d trusted it to tell me who to trust.

I still wanted human validation — real people, real stomachs, real experiences. I just didn’t want to find them, read them, weigh them myself. So I handed that to a machine. And accepted what came back as if it were the thing itself.

There’s even a concept for this — deferred trust — the mechanism by which, when evaluating other humans feels costly or uncertain, we end up delegating that decision to systems that seem neutral. Not because they’re more reliable. Because the friction is lower.

But in my case there’s something more specific happening.

I don’t trust AI in the abstract. I don’t trust it when it doesn’t show its sources. I don’t trust its reductions — the way it takes something messy and human and hands it back clean, with the uncertainty edited out. I don’t trust conclusions that arrive without the reasoning that produced them.

And yet I asked it to collect human experience on my behalf. To find the reviews, analyze them, hand me a conclusion.

Just like that. No verification. No second source. The answer arrived in a calm, confident tone, it felt coherent, and that was enough.

That’s the part I keep coming back to.

Not that I used AI. Not even that I trusted it. But that in the moment the answer arrived — I stopped asking questions.

There's a question underneath that one that I've been thinking about for a while: if we already live in a world where systems decide what's visible — what appears, what gets recommended, what enters our decision space — then the next question was always going to be this one. Not just who controls what we can find. But who shapes what we believe once we've found it.

The skyr was good.

And there’s another layer to the problem.

When the outcome is fine, there’s no rupture. No moment that forces you to examine what just happened. The delegation becomes invisible. And invisible systems don’t get examined — they just get used again.

For bigger things than yogurt.

If I trust humans more than machines — why did I let the machine decide which humans to believe?

What are we actually trusting today?

For now, just a question.

But I have the feeling that the yogurt was the easy part.

Next: Trust Is No Longer Earned.

For further reading

• Galindez-Acosta & Giraldo-Huertas, “Trust in AI emerges from distrust in humans: A machine learning study on decision-making guidance”, Universidad de La Sabana, 2025

• BCG, “Consumers Trust AI to Buy Better. Brands Need to Move Quickly”, January 2026

• Jin et al., “Algorithm Appreciation in Consumer Decisions”, International Journal of Consumer Studies, October 2025

Jimena Pérez Ferrara writes about power, technology, and the forces shaping our world — always from the human end of the story. Architect of Conversations & Voices.

Últimamente estuve pensando en lo loco que es esto de estar vivos hoy: cómo sentimos, cómo reaccionamos, cómo nos movemos en un mundo que parece cada vez más intenso y veloz.

Si uno mira las noticias, pareciera que todo está peor que nunca. Que el mundo va en una especie de espiral hacia abajo. Todo suena urgente, extremo, dramático.

Y me pregunto si eso es realmente así… o si también hay algo del algoritmo, de cómo está diseñado lo que consumimos.

¿Qué pasa cuando todo lo que vemos nos genera miedo?
¿Cómo afecta eso a nuestro sistema nervioso, a nuestras decisiones, a nuestra forma de vivir?

Porque cuando el miedo se convierte en el aire que respiramos, algo en nosotros se cierra.
Y empezamos a ver menos.

Pero al mismo tiempo, en medio de todo eso, me empezó a pasar algo.

Empecé a prestar más atención a cosas muy simples.
A interacciones humanas chiquitas, casi invisibles.

Una mirada en un negocio.
Un gesto en un aeropuerto.
Una sonrisa en la calle.

El otro día, en Waitrose, vi a una empleada pasar con uno de esos carros enormes para reponer góndolas. Otra empleada la cruzó y le dijo, así nomás : “I think that you need a hug.”

La otra la miró: ¿qué? (sinceramente yo también ahi la mire de reojo porque no sabia si había escuchado mal o que.)
Que creo que necesitas un abrazo.
Y se lo dio.

Ahí, en el medio de los yogures, las cremas y los quesos.

Me quedé mirando esa escena mínima, íntima. Y algo dentro mio se encendió, y sentí como si ese abrazo también me llegara a mí.

Y ahí entendí algo.

No es que lo malo no exista.
Pero tampoco es lo único que está pasando.

Y sin embargo, muchas veces, es lo único que vemos.

Entonces me apareció una idea muy simple:
¿y si el problema no es solo lo que pasa… sino cómo lo estamos mirando?

Porque si todo el tiempo miramos desde el miedo, desde la desconfianza, desde la idea de que todo está mal… el mundo inevitablemente se vuelve eso.

Y acá se me vino algo que tiene que ver con el momento exacto en que estamos.

Esta semana, millones de personas en el mundo están en un tiempo de pausa.
Los católicos en Cuaresma.
Los musulmanes en Ramadán.
Los judíos preparándose para Pesaj.

Tradiciones distintas, calendarios distintos, formas distintas de rezar.
Pero algo en común: un tiempo que invita a detenerse. A mirar hacia adentro. A revisar.

Y en esas tradiciones aparece algo que siempre me llamó la atención: el milagro de recuperar la vista. En distintos textos sagrados, de distintas épocas, personas que no veían… y de repente ven.

Siempre lo leí como algo literal.
Pero me pregunto si no es también una descripción de algo que puede pasarnos a cualquiera en cualquier momento.

¿Y si ese milagro también es este?

¿Y si ese milagro no viene de afuera?

¿Y si nadie viene a devolvernos la vista…
y somos nosotros los que podemos hacerlo?

No en un sentido literal, sino en algo mucho más cotidiano y, a la vez, más difícil: volver a ver.

Volver a ver a una persona sin toda la historia que le cargamos.
Volver a ver una situación sin el filtro automático del “esto está mal”.
Volver a ver nuestra vida, nuestros procesos, nuestros cambios… con un poco más de suavidad y amabilidad.

Como si, por un momento, pudiéramos salir de esa ceguera que nos hace creer que todo está perdido.

Y probar —aunque sea un poco— mirar con otros ojos.

Me pasa a veces con algo muy simple.

Veo un paisaje saliendo del invierno. Árido, marrón, seco.
Y en un primer momento pienso que feo, que no hay flores, que le falta el verde del verano, o los naranjas del otoño.
Pero si vuelvo a mirar… aparece otra cosa.

Aparece el ciclo.
Aparece que eso también es parte.
Aparece una forma distinta de belleza.

Y me pregunto si no podemos hacer lo mismo con todo.

¿Qué no estamos viendo hoy?
¿Dónde estamos siendo un poco ciegos?

No tengo una respuesta cerrada.

Pero sí tengo una invitación.

La próxima vez que sientas que todo está mal, que todo pesa, que todo es demasiado…
frená un segundo.
cerrá los ojos.
y volvé a mirar.

No para negar lo que pasa.
Sino para no perderte lo que también está pasando.

Ese pequeño ejercicio, casi invisible,
de darnos a nosotros mismos el milagro de volver a ver.

Ahí mismo.

Entre los yogures y los quesos.

-Versión en español más abajo-

I grew up in Argentina, a country where nostalgia is almost a national language.

Not nostalgia in the Instagram sense — the filtered version of the past. But the deeper kind. The kind that lives in music, in family stories, in the way older generations speak about time as if it were something that could still be touched.

In Argentina, nostalgia is almost poetic. It carries this quiet belief that somewhere, in some invisible layer of memory, things used to be better.

When I was a child, I used to come home from school and find my grandmother listening to tango. The same songs, again and again. And with them came the stories.

Stories about Buenos Aires when she first arrived from Tucuman. About cafés glowing under yellow lights. About the city when she was young, when everything seemed possible. About nights in a Buenos Aires that, in her telling, felt like it was living its own golden age.

She told those stories with such vividness that it felt cinematic. The colors of the streets, the smell of tobacco and coffee, the rhythm of a city I had never seen but could perfectly imagine.

Sometimes I wonder if it was her storytelling. Or simply my childhood ability to make those worlds real.

Tango understood this emotional terrain long before anyone tried to explain it.

One lyric captures it perfectly:

“Vivir con el alma aferrada a un dulce recuerdo que lloro otra vez.”

To live with the soul clinging to a sweet memory one mourns again.

Last month, while thinking about the strange return of corsets, early-2000s aesthetics, and the revival of 90s culture, I realized something. These weren’t just fashion cycles.

They were signals.

Signals of a culture quietly turning toward the past.

And the more I looked at the broader picture, the clearer the pattern became.

“Make America Great Again.” Russia’s narrative of restoring imperial dignity. Movements across Europe and the Global South promising to recover something — a culture, a dignity, a golden era — that was supposedly taken.

Different ideologies. Different histories. The same emotional architecture.

Nostalgia — not as sentiment, but as strategy.

Which made me wonder: why does it work so well, and why right now?

There’s research behind this — Constantine Sedikides has spent decades studying nostalgia across cultures and found something counterintuitive: it’s not really about the past. It’s about continuity.

When people experience rapid change — economic uncertainty, social fragmentation, technological disruption — nostalgia functions as a stabilizing mechanism. It restores the sense that your identity has a coherent story across time.

I am still the same person I was. The world is still recognizable.

And right now, the pace of change is extraordinary.

That speed creates something closer to a nervous system response.

When the system is overwhelmed, humans reach for what is familiar. Something stable. Something known.

Children do this naturally. When they feel uncertain, they hold something — a toy, a blanket — something that says the world is still safe.

Could nostalgia be doing the same thing for entire societies?

A collective attachment object. Psychological infrastructure, not just sentiment.

Think about what you’ve been reaching for lately. A playlist from ten years ago. A show you already watched. A style that reminds you of something. None of it is random.

You don’t have to look far to see it in action.

This week, feeds everywhere filled with childhood photos. People in their 20s, 30s, 40s posting a picture of themselves at age 6 with the caption: “met my team” — one kid, scraped knees, zero agenda. Or a current photo on top, childhood photo below: “this is who’s behind the business.”

It’s not a random trend. It’s a collective response to exactly this. When the present feels too fast, too heavy — we go looking for the kid we were. As if to say: there was a version of me before all of this. And she’s fine.

What does it tell us that the most viral content right now is, literally, the past?

A 2025 Hootsuite report found that nostalgia-driven posts get 30% more shares and likes than regular content. The platforms know.

And here’s the uncomfortable follow-up: what happens when the systems we use every day start to notice that pattern — and feed it back to us?

Byung-Chul Han writes about cultural atomization — shared symbolic frameworks dissolving into highly individualized identity loops. Fragmentation doesn’t just weaken common narratives. It creates openings.

When those openings appear, algorithms move in. Recommendation systems are optimized for engagement, and engagement often favors the familiar. Generative AI reproduces aesthetic signatures from the past. The feed learns what you long for — and then decides how much of it to give you.

Researchers are already calling it “algorithmic nostalgia” — the idea that platforms don’t just reflect what we long for, they actively produce it.

The question I keep sitting with: are algorithms detecting nostalgia in us, or are they producing it?

Because there’s a difference between a tool that reflects what we already long for — and a tool that slowly teaches us what to long for.

And some actors have become extremely effective at exploiting those openings. Not only political movements. Platforms themselves.

Which brings a harder question: whose nostalgia gets encoded?

If the majority of training data is Western and English-dominant — whose memory becomes the default? Which golden eras get amplified, and which ones quietly disappear? You probably can’t feel it happening. That’s the point.

Watch what happened with K-pop and Korean skincare. What started as cultural exports became a deliberate soft power strategy — South Korea embedding its presence globally through aesthetics, music, routines, digital communities. Joseph Nye wrote about soft power as the ability to attract rather than coerce. South Korea did exactly that, but without governments leading — with fandoms, with algorithms, with platforms, with the quiet ritual of a 10-step skincare routine that suddenly everyone had an opinion about.

The difference is that South Korea chose to encode its own memory. Most of the world didn’t get that choice.

I often return to those afternoons listening to my grandmother’s stories.

Nostalgia is powerful because it creates continuity. It allows people to believe that the past, the present, and the future belong to the same narrative.

But nostalgia can also become a trap.

Societies that only look backward eventually lose their ability to imagine what comes next.

What does it mean to use memory without being captured by it?

Maybe that’s why one of the most beautiful lines in Argentine music refuses nostalgia altogether

As Luis Alberto Spinetta once wrote:

“Aunque me fuercen yo nunca voy a decir que todo tiempo por pasado fue mejor… mañana es mejor.”

No matter how hard they push me, I will never say that the past was better. Tomorrow is better.

Not because tomorrow is guaranteed.

Even on days when I doubt humanity, something in me still believes we are capable of rebuilding. Of reinventing ourselves. Of choosing empathy and care, even when everything seems to say otherwise.

And maybe that quiet belief is what keeps tomorrow open.

Because as my grandmother used to say — hope is the last thing you lose.

What are you noticing about the nostalgia around you? Is it comfort, or is it strategy — and can you tell the difference?

La Máquina de la Nostalgia.

Soy Argentina, un país donde la nostalgia es casi un idioma nacional.
No la nostalgia de Instagram, la versión filtrada del pasado. Sino la más profunda. La que vive en la música, en las historias familiares, en la manera en que los abuelos hablan del tiempo como si todavía se pudiera tocar.

En Argentina, la nostalgia es casi poética. Carga esa creencia silenciosa de que en algún lugar, en alguna capa invisible de la memoria, las cosas fueron mejores.

De chica, llegaba del colegio y encontraba a mi abuela escuchando tango. Las mismas canciones, una y otra vez. Y con ellas llegaban las historias.
Historias sobre Buenos Aires cuando ella llegó desde Tucumán. Sobre la ciudad cuando era joven, cuando todo parecía posible. Sobre noches en un Buenos Aires que, en su relato, vivía su propia época dorada.

Las contaba con tanta nitidez que parecían películas. Los colores de las calles, el olor a tabaco y café, el ritmo de una ciudad que yo nunca había visto pero podía imaginar perfectamente. A veces me pregunto si era su forma de contar o mi capacidad infantil de hacer real lo que me relataba.

El tango entendió este territorio emocional mucho antes de que alguien intentara explicarlo. Una letra lo captura perfecto:
“Vivir con el alma aferrada a un dulce recuerdo que lloro otra vez.”
La nostalgia, en ese sentido, no es solo recordar. Es una forma de sostener la identidad a través del tiempo.

El mes pasado, pensando en el regreso extraño de los corsés, la estética de los 2000 y el revival de los 90, me di cuenta de algo: no eran simples ciclos de moda. Eran señales. Señales de una cultura que se giraba, en silencio, hacia el pasado.

Y cuanto más miraba el panorama, más claro se volvía el patrón:
“Make America Great Again.” La narrativa rusa de restaurar la dignidad imperial. Movimientos en Europa y el Sur Global prometiendo recuperar algo —una cultura, una dignidad, una época dorada— que supuestamente les fue arrebatado.

Ideologías distintas. Historias distintas. La misma arquitectura emocional.

La investigación respalda esto. Constantine Sedikides estudió la nostalgia en distintas culturas y encontró algo contraintuitivo: no se trata realmente del pasado. Se trata de continuidad.
Cuando vivimos cambios acelerados —incertidumbre económica, fragmentación social, disrupción tecnológica— la nostalgia actúa como mecanismo estabilizador. Restaura la sensación de que nuestra identidad tiene una historia coherente.

Sigo siendo la misma persona que era. El mundo sigue siendo reconocible.
Y hoy, el ritmo del cambio es extraordinario. Esa velocidad genera algo cercano a una respuesta del sistema nervioso. Cuando se satura, buscamos lo familiar. Algo estable. Algo conocido.

Los niños lo hacen naturalmente. Cuando sienten incertidumbre, se aferran a algo: un juguete, una manta, algo que les dice que el mundo sigue siendo seguro.

¿Podría la nostalgia estar haciendo lo mismo para sociedades enteras? Un objeto de apego colectivo. Infraestructura psicológica, no solo sentimiento.

Pensá en lo que buscaste últimamente: una playlist de hace diez años, una serie que ya viste, un estilo que te recuerda a algo. Nada de eso es azar.

Esta semana, los feeds se llenaron de fotos de infancia. Personas de 20, 30, 40 años posteando fotos de cuando tenían 6 con el caption: “conocé a mi equipo” O “ella es la que se encarga de revisar los pedidos “.

No es una tendencia aleatoria. Es una respuesta colectiva. Cuando el presente se siente demasiado rápido, salimos a buscar al niño que fuimos. Como diciendo: hubo una versión de mí antes de todo esto. Y está bien.

Un informe de Hootsuite de 2025 encontró que los posts con nostalgia generan un 30% más de shares y likes que el contenido regular. Las plataformas lo saben.

Y acá viene la pregunta incómoda: ¿qué pasa cuando los sistemas que usamos todos los días notan ese patrón y nos lo devuelven amplificado?
Byung-Chul Han habla de atomización cultural: marcos simbólicos compartidos que se disuelven en loops de identidad individualizados. La fragmentación no solo debilita relatos comunes. Crea aperturas.

Cuando esas aperturas aparecen, entran los algoritmos. Los sistemas de recomendación priorizan lo familiar. La IA generativa reproduce firmas estéticas del pasado. El feed aprende lo que añorás, y luego decide cuánto darte.

Algunos investigadores llaman a esto “nostalgia algorítmica”: las plataformas no solo reflejan lo que anhelamos, lo producen activamente.

La pregunta que me surge es: ¿los algoritmos detectan la nostalgia o la fabrican?

Mirá lo que pasó con el K-pop y el skincare coreano. Lo que empezó como exportación cultural se volvió una estrategia deliberada de soft power: Corea del Sur incrustando su presencia global a través de estética, música y comunidades digitales.

El soft power ya no se mueve principalmente a través de instituciones. Se mueve a través del código.

Vuelvo a esas tardes escuchando a mi abuela. La nostalgia crea continuidad. Permite creer que pasado, presente y futuro pertenecen al mismo relato. Pero también puede convertirse en trampa. Las sociedades que solo miran atrás pierden la capacidad de imaginar lo que viene.

Como escribió Luis Alberto Spinetta:
”Aunque me fuercen yo nunca voy a decir que todo tiempo por pasado fue mejor… mañana es mejor.”

Incluso en los días en que dudo de la humanidad, algo en mí sigue creyendo que somos capaces de reconstruirnos De reinventarnos.
De elegir la empatía y el cuidado, incluso cuando todo parece decir lo contrario.

Y quizás esa creencia silenciosa es lo que mantiene el mañana abierto.

Por que como decía mi abuela, la esperanza es lo ultimo que se pierde.

This essay is about power, agentic AI, and what I'm calling Discovery Sovereignty.

A sentence I read recently in a LinkedIn post by Dr. Nicolai Schättgen has stayed with me:

“If the agent can’t discover you, you don’t exist to it.”

He was writing about AI infrastructure. About protocols. About the invisible architecture of how systems navigate the world.

It’s a technical observation. But the moment I read it, something clicked. Not because it was unfamiliar — because it was the opposite. I studied International Relations. I spent years reading about how power moves through networks, how the architecture of systems determines who gets access and who gets cut off. This sentence wasn’t introducing a new concept. It was describing something I had studied in a completely different context, showing up in a completely different layer.

The same logic. Just migrated.

The pattern behind the headlines

The world is in visible fracture. Russia’s full-scale invasion of Ukraine in 2022 shattered European security assumptions that had held since the end of the Cold War. Escalating tensions between the United States and Iran have repeatedly brought the region to the brink of direct confrontation. In parallel, recent U.S. administrations have shown a growing willingness to instrumentalize alliances and economic interdependence in ways that have unsettled even long-standing partners.

In this context, Europe’s renewed conversation about nuclear deterrence, independent defense capacity, digital sovereignty, and strategic autonomy isn’t panic.

It is infrastructure planning.

To understand why, you need one idea from political scientists Henry Farrell and Abraham Newman: weaponized interdependence. Their argument, developed across a decade of work and updated most recently in Foreign Affairs, starts from something structural. Global networks were built as neutral conduits — financial systems, internet infrastructure, supply chains, all celebrated as the architecture of mutual benefit. What this missed is that networks never stay flat. They centralize. They produce hubs. And hubs are never neutral.

Whoever controls the hub can do two things: see everything that flows through it — the panopticon effect — and cut off whoever they want — the chokepoint effect. The United States exercised both for decades. When Iranian banks were disconnected from the global financial messaging system SWIFT in 2012, it wasn’t merely a financial sanction. It was the economic equivalent of a siege.

Macron, Poland, the Baltic states aren’t reacting to a single event. They’re responding to a structural realization: the hub they built everything on — American security guarantees, American technology, American-led institutions — may no longer be politically neutral or fully predictable. When the hub becomes conditional, you don’t deliberate abstractly. You build redundancy. What sounds like a conversation about weapons is actually a conversation about dependency: which layer do you control, and which are you renting from someone who could revoke access?

The layer nobody is watching

The same dynamic is playing out at a different scale — and almost nobody is reading it as the same story.

Every major organization right now is asking a quieter version of the same question. The company that built its customer acquisition on a platform algorithm discovered, when the algorithm changed, that it never owned the layer. The enterprise that migrated entirely to a single cloud provider is asking what data sovereignty means when your operations run on someone else’s infrastructure, under someone else’s jurisdiction. The media brand that built its audience inside a social network watched that audience become unreachable overnight.

Consulting firms are building platforms. Tech giants are selling strategy. The lines have dissolved — because control of the layer, not production of the product, is where the value lives.

Benjamin Bratton saw this coming from a different angle. In The Stack: On Software and Sovereignty, he argued that computational infrastructure doesn’t just reflect political geography. It produces it. Whoever designs the layers of how information flows defines what can exist and what disappears. Sovereignty, he proposed, has migrated from territory to stack.

The rise of agentic AI adds a further shift. An AI agent doesn’t simply search — it navigates, synthesizes, and decides on your behalf what exists in your decision space. The infrastructure of discovery is no longer passive. It becomes an active participant in determining what is findable, what is valuable, what gets recommended.

This is where both frameworks converge.

The discovery layer of agentic AI is emerging as a new hub. Whoever controls the protocols through which agents find information may hold both panopticon and chokepoint power over the knowledge economy now being built.

Three scales, one logic

Here is what I find most striking about this moment: the macro, the meso, and the micro are the same structural problem wearing different clothes.

And we almost never talk about them in the same room.

At the macro scale, the geopolitical fracture isn’t primarily about ideology or personality. It’s structural. States that spent decades building their security, economy, and technological future on top of a hub they didn’t control are now confronting the implications of that dependency. Strategic autonomy is not rhetorical. It is infrastructural.

At the meso scale, the same realization is hitting organizations. The boardroom question — which cloud provider, which platform, which algorithm, which AI system — is structurally identical to the question European leaders are asking about defense and digital sovereignty. A company that built its entire customer acquisition on a single platform algorithm discovered, when that algorithm changed, that it never owned the layer. It owned the output. The layer belonged to someone else.

At the micro scale, the same logic lands on individuals. If an AI agent becomes the first filter between a potential client and the professionals they might hire, then your visibility within that agent’s discovery layer is not simply a personal branding issue. It is an access issue. It determines whether you enter the decision space at all.

You can be brilliant. You can have the right experience, the right network, the right offer. And if the system that mediates access cannot find you — if the agent cannot discover you — none of it gets a chance to matter.

The chokepoint effect, operating at the scale of a single person.

What connects all three scales is the same underlying shift: the entity that controls discovery increasingly controls relevance. This is what I'd call Discovery Sovereignty — and it operates at every scale simultaneously.

Farrell and Newman argued that economic coercion follows the hub. Bratton has argued since 2015 that sovereignty follows infrastructure. What this moment suggests is that sovereignty is extending into the architecture of discovery itself.

Power has never only been about who produces the most, who knows the most, or who has the loudest voice.

It has always been about who controls access to what matters.

That question is being asked and answered right now, simultaneously, at every scale. States are rebuilding security and digital infrastructure because they realized their hub was conditional. Organizations are auditing their dependencies because they realized their layers were rented. And individuals — professionals, experts, builders — are discovering that visibility in the systems that mediate access is no longer optional. It is a structural condition.

The geopolitical fracture, the digital sovereignty debate, and the agentic discovery problem are not three separate stories. They are three expressions of one dynamic playing out at different scales, at the same time. The entities that understand this now will be positioned to navigate what comes next. Those that don’t will find themselves not defeated but invisible.

We spent a decade debating digital sovereignty. We may be entering the era of Discovery Sovereignty — where what you can be found determines what you can become.

Power has always lived in the layer nobody watches.

The layer just changed.

The question — which layer do you control, and who can find you there? — has not.

Jimena Pérez Ferrara writes about power, technology, and the forces shaping our world — always from the human end of the story.

For further reading

• Henry Farrell & Abraham Newman, “The Weaponized World Economy”, Foreign Affairs, September/October 2025

• Henry Farrell & Abraham Newman, Underground Empire: How America Weaponized the World Economy, Henry Holt, 2023

• Henry Farrell & Abraham Newman, “Weaponized Interdependence: How Global Economic Networks Shape State Coercion”, International Security, Vol. 44, No. 1, 2019

• Benjamin H. Bratton, The Stack: On Software and Sovereignty, MIT Press, 2015 (10th anniversary edition, 2026)

Y sin pensarlo, ya estamos en marzo.

La sensación de que febrero fue eterno está instalada en mi cuerpo.

¿De verdad todo eso pasó en un solo mes?

¿O fue hace mil días?

Más que los hechos, me impresiona la intensidad. Todo en lo que me sumergí. Todo lo que, aunque fuera por una semana, hice mío.

Hoy, 3/3, miro hacia atrás y pienso: wow.
¿Es posible que todo eso haya coexistido?
¿Fue solo algo mío o nos pasó a todos?
¿Me subí a todos los trends solo porque me atrapaban?

Pero después de las preguntas aparece otra capa:

¿Cuál es el hilo conductor?
¿Qué estamos buscando las personas hoy?

• ¿Será sentirnos más humanos?

• ¿Reconocernos en historias de pasiones extremas, memorias y orgullo colectivo?

• ¿Será nostalgia?

• ¿Será pertenencia?

No lo sé. A mí las preguntas me gustan abiertas, respirando, sin una sola respuesta.

Cumbres Borrascosas. El corset. Jacob. Los paisajes. El amor y el odio en su versión más cruda. Esa estética que exige postura, que aprieta, que incomoda.

¿Por qué nos atrae lo que nos tensa?
¿Hay algo en lo visceral, en lo 100% humano, que casi 180 años después vuelve a sentirse urgente?
¿Será que cambian los contextos pero no las emociones?
¿Que seguimos siendo los mismos cuando amamos, odiamos, deseamos?

El latinaje. El perreo de Bad Bunny en el Super Bowl.

De repente, América Latina mainstream.

Una performance artística que también era un estandarte político. Sin discurso explícito. Sin mencionar lo que había pasado cinco minutos antes en Minnesota. Pero dejando algo claro:

El poder simbólico se mueve. Las narrativas cambian.

Hay un término para esto: soft power. La capacidad de influir sin coerción, a través de cultura, valores, narrativa. Lo teorizó Joseph Nye para hablar de geopolítica entre estados. Pero Bad Bunny lo hizo esa noche con un corte de pelo y canciones en español en el escenario más visto del año. La periferia hablándole al centro sin pedir permiso. Sin traducirse.

Y yo pensando: aguante América Latina, aguante todo.

Lo que parecía solemnidad se mezcla con ritmo. Lo que parecía estructura se vuelve movimiento.

¿Cómo conviven rigidez y liberación?
¿Cómo transformamos identidad en presencia?
¿Cuándo el orgullo deja de ser defensivo y se vuelve celebración?

Y como si no fuera suficiente, aparece Love Story.

Los 90. La música que escuchaba mi mamá.
Mi abuela hablando de JFK Jr., recordando que tuvo un perrito llamado John John.

Una estética distinta: la moda de los 90, minimalista y NET, donde lo masculino y lo femenino se cruzaban y cada prenda contaba un gesto de identidad.

Nostalgia y presente bailando juntos.

El pasado vuelve, pero editado, reinterpretado. ¿Estamos recordando o reescribiendo?
¿Adoptamos lo que nos conviene del pasado?

¿Cuál es el hilo?

Y cuando parecía que febrero había terminado, el último día termina con una guerra.

Una guerra que parecía cantada, pero se siente como portal.

No sé si algo del sistema internacional está mutando, pero lo siento así.

Y otra vez: no tengo respuestas. Solo preguntas.

¿Cómo se redefine el poder en un mundo donde conviven corsets victorianos, perreo global y conflictos geopolíticos en tiempo real?
¿Cómo diferenciamos lo estructural del ruido?
¿Cómo no anestesiarnos frente a tanto estímulo?

Al final, el hilo conductor no es un hecho, ni un estilo, ni un evento.

Es la capacidad de conectar experiencias humanas que parecen dispares y descubrir que no lo son tanto.

Es preguntarse, más que afirmar:

¿Qué me mueve?
¿Qué resuena conmigo?
¿Qué es lo que importa y qué es solo confetti?

Febrero fue un collage. Un ensayo de vida.
Corsets y perreo. Nostalgia y guerra. Estética y cuerpo. Espectáculo y realidad.

Y mi poder —si es que puedo llamarlo así— es unirlo, sentirlo, interrogarlo y compartirlo, siempre desde la mirada que observa.

Porque quizá, al final, lo único que podemos hacer es preguntar, sentir y sostener la curiosidad infinita mientras el mundo gira y cada experiencia nos devuelve un eco de lo que somos.

Hay rituales que uno no cuestiona.
Desde hace años, cada comienzo de año espero el reporte de tendencias de VML.

No lo leo solo para “ver qué viene”.
Lo leo para algo más íntimo:
para inspirarme, para calibrar si mi brújula interna sigue orientada,
para detectar qué nuevas ideas me dan ganas de aprender con más profundidad.
Y, sobre todo, para leer entre líneas qué nos está pasando como humanos.

Porque más allá de cualquier hype, cualquier tecnología o cualquier crisis,
seguimos siendo personas tratando de entender el mundo que habitamos.

Este año, The Future 100: 2026 puso en palabras algo que venía sintiendo hace tiempo.
Un concepto que no promete luz permanente ni niega la oscuridad.

Lo llamaron Dystoptimism.

Un neologismo que combina distopía y optimismo.
Reconocer que el mundo es complejo, frágil y a veces oscuro…
sin rendirse ante eso.
Encontrando posibilidad en la renovación.

Cuando leí la palabra, mi cabeza se fue directo a la idea clásica de lo distópico.
El cuento de la criada.
Ese futuro que antes parecía lejano, exagerado, casi de ficción.

Y entonces la pregunta incómoda apareció sola:
¿acaso ese “futuro distópico” no es, en parte, el presente que ya estamos viviendo?

Las noticias que vemos todos los días.
La coyuntura global.
Lo que sucede en lugares que nos quedan lejos… y lo que ni siquiera sabemos que ya está pasando detrás del escenario.

El mundo no espera.
Nunca esperó.

Tal vez por mi background —sí, quizás porque soy argentina—
siempre sentí que la adaptabilidad, la flexibilidad y la capacidad de reinventarse no eran una opción, sino una condición de supervivencia.

Y ahí es donde este concepto empezó a resonar más hondo.

Porque Dystoptimism no niega la oscuridad,
pero tampoco se queda paralizado en ella.

Como explican Emma Chiu y Marie Stafford, Global Directors de VML Intelligence:

“Dystoptimism destaca que cuando los sistemas antiguos se desmoronan, individuos, comunidades e innovadores están construyendo nuevas soluciones centradas en el ser humano.”

No se trata de idealizar el pasado.
Se trata de diseñar activamente algo nuevo mientras todo se mueve.

El propio reporte muestra un dato que me pareció clave:

→ 86% de las personas buscan experiencias que renueven su visión del mundo,
que generen asombro, perspectiva y sentido.

No evasión.
No optimismo vacío.
Asombro como forma de volver a mirar.

Y eso me hizo pensar en algo más amplio que el consumo, el trabajo o las tendencias.
Me hizo pensar en cómo estamos eligiendo habitar la incertidumbre.

Ya no alcanza con negar lo difícil.
Pero tampoco con quedar atrapados en el miedo.

Lo que empieza a emerger es otra cosa:
una forma de estar en el mundo que acepta la complejidad
y, aun así, busca colaboración, humanidad y creación colectiva.

Quizás eso sea lo verdaderamente disruptivo hoy.
No la tecnología en sí, sino qué hacemos con ella.
No los cambios, sino desde dónde los atravesamos.

Dystoptimism no es una palabra de moda.
Es el nombre de un estado mental que ya está entre nosotros.

Y, tal vez, una invitación silenciosa a algo más profundo:
seguir mirando de frente lo que duele,
sin dejar de construir lo que todavía puede florecer.

— Jimena.

Hoy, 6 de agosto, una semana después vuelvo, y si tuviera que escribir todo lo que pasó, quizá debería escribir un libro. Solo lo voy a simplificar como los regalos que tiene la vida cuando uno se mantiene flexible ante los planes que, seguramente, son mejores que los que tenemos en nuestra mente.

Así fue como recibí un regalo inesperado: un fin de semana en una dairy farm en Newbury, en el corazón del hermoso campo inglés. Estuve con Lockie, Pucci, y nuevos amigos en una barbacoa, disfrutando del placer de descansar en la tranquilidad de la naturaleza. La amabilidad de estas personas me hizo sentir en casa en solo unas horas. Es mágico cómo el amor puede fluir entre extraños que se convierten en amigos cercanos, y cuán reconfortante puede ser esa conexión.

Y luego el movimiento nuevamente, ¿acaso se siente la vida como el océano? Por momentos, el mar en calma en un día de sol tibio, la capacidad de tomar aire, respirar profundo, disfrutar del éxtasis de la vida y la gratitud infinita. Y en otros momentos, el movimiento, las olas, el sentimiento de no tener una base firme. En esos momentos de inestabilidad, respiro y pienso: “Ok, ¿qué es lo próximo chiquito que necesito para poder sostenerme? ¿Cómo me puedo sostener a mí misma? ¿Qué es lo que me puede hacer sentir bien y empezar a recuperar la calma? ¿Cómo puedo tomar perspectiva en mi vida y realmente sentirme agradecida?” Y todo pasa, y los sentimientos se calman, y la vida se vuelve a sentir como un mar en calma.

Cuando me tomé el momento de conversar sobre mis sentimientos y pude poner en dimensión la situación, comprendí que un mal día no es una mala vida, y luego me corrigieron y la verdadera lección fue que inconclusive un mal momento del día no es un mal día, y nuevamente todo toma otra perspectiva.

Encontrar la belleza en cada día, encontrar el placer en cada día, hacer de cada día un ritual sagrado y sostener la gratitud infinita por el simple hecho de tener un día más de vida.

Y en esta búsqueda, comencé algo que llamo "Bitácora del Placer". Enumero cosas en el día que me dan placer en un pequeño cuaderno. Hace días que no escribía nada y hoy volví a escribir:

#36 Encontrar algo preciado que pensé que había perdido.

En este caso, fue mi amada piedra turmalina, que siempre llevo conmigo como amuleto de protección. Hacía días que no la veía y pensé que estaba perdida. Hoy apareció en un pantalón que me puse.

Así es la vida. Encontrar algo preciado que pensé que había perdido, volver a ver la vida con asombro, encontrar la belleza en todo y en todos, y sobre todo en momentos de movimiento exterior, encontrar la calma, la confianza y la certeza en mi interior.

Este mélange de escritura, como el mélange de la vida, sus matices, sus colores y la belleza de un sillón de cada color. Hasta hace minutos, me parecía que debían estar todos con el mismo tapizado, a lo sumo engamados. Levanto la vista, los vuelvo a ver y solo puedo apreciar su belleza, su autenticidad y lo mágica y sincrónica que es la vida.

What am I going to write about? This page says 30/7, which was when I decided to write again, but it was only because a week had passed since the last entry, and I assumed I should do it. I didn’t feel it, so I just wrote the date. Then, I remember I got up and went to read.

Today, August 6th, a week later, I come back, and if I had to write everything that happened, I might need to write a book. I’ll just simplify it as the gifts life offers when we stay flexible with plans that are surely better than those we have in our minds.

And so it was that I received an unexpected gift: a weekend on a dairy farm in Newbury, in the heart of the beautiful English countryside. I spent time with Lockie, Pucci, and new friends at a barbecue, enjoying the pleasure of resting in the tranquility of nature. The kindness of these people made me feel at home in just a few hours. It’s magical how love can flow between strangers who become close friends, and how comforting that connection can be.

And then, the movement again. Does life feel like the ocean? At times, the calm sea on a warm sunny day, the ability to take a breath, breathe deeply, enjoy the ecstasy of life, and infinite gratitude. And at other times, the movement, the waves, the feeling of not having a firm base. In those moments of instability, I breathe and think: “Okay, what’s the next little thing I need to hold on? How can I support myself? What can make me feel good and start to regain calm? How can I gain perspective on my life and truly feel grateful?” And everything passes, the feelings calm down, and life starts to feel like a calm sea again.

When I took the time to talk about my feelings and put the situation into perspective, I realized that a bad day is not a bad life, and then I was corrected. The real lesson was that an unfortunate moment of the day is not a bad day, and once again, everything takes on a new perspective.

Finding beauty in each day, finding pleasure in each day, making each day a sacred ritual, and holding onto infinite gratitude for the simple fact of having one more day of life.

And in this quest, I started something called “Pleasure Log.” I list things each day that bring me pleasure in a small notebook. I hadn’t written anything in days, and today I wrote again:

#36 Finding something precious that I thought I had lost.

In this case, it was my beloved tourmaline stone, which I always carry with me as a protective amulet. I hadn’t seen it for days and thought it was lost. Today, it appeared in a pair of pants I put on.

That’s life. Finding something precious that I thought I had lost, seeing life with wonder again, finding beauty in everything and everyone, and especially in moments of external movement, finding calm, confidence, and certainty within myself.

This mélange of writing, like the mélange of life, its nuances, its colors, and the beauty of a sofa in every color. Until moments ago, it seemed to me that they should all have the same upholstery, at most coordinated. I lift my gaze, look at them again, and can only appreciate their beauty, their authenticity, and the magical and synchronous nature of life.

La Belleza en el Melange - The Beauty in the Mélange

Mi preciada Turmalina. - My Precious Tourmaline

Desayuno con Pucci en Dairy Farm House -Marsh Benham Newbury. - Breakfast with Pucci at Dairy Farm House - Marsh Benham Newbury

Cuanta belleza en la contemplación de la naturaleza, cuanta belleza en las flores, siempre me gustaron, siempre creí en el poder de la belleza de la naturaleza y en su sabiduría infinita, en comprender sus ciclos, lo impermanente de las rosas, pero a su vez cuán magníficas se vuelven cuando florecen, sus formas, su aroma, sus espinas su textura media pegajosa de las hojas y los tallos cuando queres cortar la rosa. Y así la vida misma.

295 días en los cuales solté lo que pensé que era hogar, 295 días en que mis metas fueron cambiando, hasta hoy que escuche vos sos la meta, y caló hondo, como yo soy la meta? si tengo miles de metas, y así quedó resonando en mi cabeza, vos sos la meta. Y si yo soy la meta? si todo esto es realmente lo mágico de esta experiencia humana, si cada uno es la meta, ¿Cómo deberíamos vivir? ¿Cómo podría honrar esa meta? solo preguntas y más preguntas. Que hermosas son las preguntas.

Este experimento de escritura, arrancó hoy 23 de Julio del 2024, quizás dure una semana, quizás no vuelva a escribir o le tomé el gusto a escribir, no lo sé. Lo que si se es que voy a tratar de no editarme, simplemente el ejercicio es escribir lo que sienta, lo que piense, sin editarme, ni editar el texto. A veces pienso quien va a leer esto, lo escribo para que alguien lo lea? lo escribo para mi? para que lo escribo? voy a monetizarlo? ¿Estoy perdiendo el tiempo? y luego vuelvo, respiro y me digo que importa? por ahora solo escribo, ya será el tiempo de las respuestas.

295 Days

295 days ago, I left what I thought was my home. 289 days ago, I left the city where I had lived for the past 19 years. I’d like to think and remember what that 17-year-old Jimena thought when she arrived in the city to study: what her dreams were, why she chose that career, and why she always knew she had to go to Buenos Aires. How many lives lived in those 19 years, but the most magnificent thing that comes to mind is how many lives I’ve lived in these past almost 300 days, how many times I’ve changed, how many parts of me I had to let die, how many times I resurrected and was reborn. How everything changed.

At times, it felt like a samba; other times, like a roller coaster, the kind that gives you the most fear. And at other times, it felt quite similar to life itself. The beautiful, the sharp, the tangled, the beauty of chaos, the ability to be reborn, the opportunity to re-question, and simply stay with the questions without needing the answers. Because, as I learned once, some years are for questions and other years are for answers. And that’s the perfection of life: to admire the question, to ask good questions, to capture in the question the capacity to reflect and admire the beauty of chaos, and above all, to have patience. Because knowing how to ask good questions not only requires introspection and absolute presence, but also the ability to sustain silence, to sustain uncertainty, and to let the miracle of life appear in that infinite stillness — the blessings, the unexpected — or always expected in even more magical and wonderful ways than our minds could imagine.

How much beauty in the contemplation of nature, how much beauty in flowers. I have always liked them; I always believed in the power of the beauty of nature and its infinite wisdom, in understanding its cycles, the impermanence of roses, but at the same time how magnificent they become when they bloom: their shapes, their aroma, their thorns, the slightly sticky texture of the leaves and stems when you try to cut the rose. And so it is with life itself.

295 days in which I let go of what I thought was home, 295 days in which my goals have been changing, until today, when I heard, “You are the goal,” and it hit deep. How am I the goal? I have thousands of goals, and it kept resonating in my head: “You are the goal.” And if I am the goal? If all this is truly the magic of this human experience, if each of us is the goal, how should we live? How could I honor that goal? Just questions and more questions. How beautiful are the questions.

This writing experiment started today, July 23, 2024. It might last a week; maybe I won’t write again, or maybe I’ll take a liking to writing. I don’t know. What I do know is that I will try not to edit myself; the exercise is simply to write what I feel, and what I think, without editing myself or the text. Sometimes I think, who will read this? Am I writing for someone to read it? Am I writing for myself? Why am I writing it? Will I monetize it? Am I wasting my time? And then I return, breathe, and tell myself, what does it matter? For now, I just write. The time for answers will come.

Subscribe now