When AI Chooses What We Get to Choose

AI promises to free travelers from generic recommendations, crowded destinations, and irrelevant choices. Yet travel reveals a broader shift: personalized systems do more than help us choose; they shape what enters our field of vision. Whether guided by inferred behavior or declared preference, they may leave us with fewer common experiences—and fewer unchosen encounters that might change who we are.

William Powell Frith, The Railway Station (1862), oil on canvas. Royal Holloway, University of London. Public domain, via Wikimedia Commons.

By Martina Moneke

The Economist recently considered a quiet irony of modern travel: artificial intelligence promises a tailor-made journey just as tourism’s viral popularity funnels travelers toward the same suffocating choke points. Rather than directing everyone toward crowded Pompeii, Santorini, or Kyoto in peak season, an AI adviser might suggest the quieter ruin, the overlooked neighborhood, or the obscure museum suited to an individual traveler.

The appeal is undeniable. Personalized planning may disperse destructive concentrations of tourism, direct visitors and spending beyond famous destinations, accommodate complex accessibility needs, and strip away hours of irrelevant searching. It offers a potential counterweight to the commercial uniformity produced by review rankings, social feeds, and top-ten lists.

Yet the article identifies a subtler cost: in abandoning the beaten track, travelers may also abandon the shared experiences that form around it. The point is not that famous sites are inherently superior or that overcrowding is a civic virtue. Rather, travel exposes a broader transformation. The central question is no longer merely which path we choose through the world; it is how the world becomes visible enough to be chosen at all. Travel gives physical form to an architecture that increasingly organizes attention everywhere else—from the music we stream and the news we read to the political arguments we encounter.

To understand what is changing, we must clear away a comforting myth: there was never an innocent age of unmediated encounter. Every era’s choices have been conditioned by prior selections. Guidebooks, museum directors, newspaper editors, state curricula, transport networks, prices, and geographic borders have always shaped what became visible and accessible. The shift we are living through is not from uncurated freedom to algorithmic control. It is the quiet replacement of broadly shared fields of curation with increasingly individualized ones.

A traditional guidebook, public broadcasting schedule, or morning newspaper may be deeply partial, but it presents a substantially common selection to a broad audience. Different people encounter the same front page, exhibition wall, or recommended itinerary; they may embrace, criticize, or ignore its contents, but they do so within a shared frame. A personalized system performs an earlier, less visible act. It adjusts which possibilities enter each person’s field of vision, calibrating visibility to a prior account of who that person is.

The mechanisms vary. Recommender algorithms may infer preference from clicks, watch time, purchases, and demographic proxies. A conversational chatbot may instead respond to preferences explicitly declared by a user. Some systems may combine inferred, remembered, contextual, and declared information. But the fundamental issue does not depend on covert surveillance or behavioral manipulation. Even when transparent and user-directed, personalization organizes possible experience around a prior account of what the person considers relevant.

The result is a subtle algorithmic provincialism. A personalized system may offer thousands of options yet repeatedly return us to the province of the already relevant. A traveler asking for historic churches may receive a brilliant array of them without ever encountering the neighborhood festival, local theater, or industrial ruin they did not know enough to request. The problem is not that experience has been curated, but that it has been curated according to a model of the person encountering it.

A common culture does not require everyone to admire the same monuments, read the same books, or hold identical political views. Democratic cohesion has never demanded uniform agreement. Rather, strangers need enough objects in common to sustain interpretation, debate, and disagreement across different positions.

A widely read novel, public memorial, shared holiday, major broadcast, or national election can furnish reference points through which private lives connect. Such objects generate overlapping memories, metaphors, evidence, and arguments. In The Human Condition, Hannah Arendt likened the “common world” to a table around which people sit: it simultaneously relates and separates them, giving distinct perspectives something upon which to converge. In this sense, common culture functions as civic infrastructure. Just as roads connect places, shared cultural references make private interpretations mutually intelligible.

The danger of pervasive personalization is not merely that it may deepen ideological polarization, but that it may gradually thin the shared object about which disagreement occurs. As feeds, itineraries, and news digests are increasingly customized around individualized accounts of relevance, the common frame can begin to shrink. A tourist queue or viral café is not inherently an arena of democratic deliberation. But the article’s observation about the crowd points toward a deeper vulnerability: when fewer experiences and objects remain broadly shared, the common world grows more fragile. A society divided by conflicting opinions can still deliberate. A society whose members are directed toward different objects according to personalized accounts of relevance may struggle even to establish what its disagreements concern.

That threat reaches inward as well, altering the conditions under which an individual can change.

The logic is understandable: what a person has previously valued—or presently says they value—is useful evidence of what might suit them next. But human preferences are not static needs awaiting accurate fulfillment. Art, travel, education, friendship, and democratic debate do not merely satisfy desires; at their most consequential, they transform them.

A preference-satisfying experience gives us more of what we already want. A preference-transforming experience alters what we are capable of wanting. Some of the encounters that change us most arrive unbidden—appearing initially irrelevant, difficult, foreign, or outside the vocabulary we use to describe ourselves. We may seek transformation, but we cannot always specify the experience that will produce it, because its value becomes intelligible only after it has begun to reshape us.

This is where friction enters, but only as a subordinate element. A wrong turn, an unplanned delay, or proximity to strangers can interrupt self-enclosure. Yet inconvenience is not inherently noble, and difficulty should not be romanticized. A system could recommend challenging art, exhausting travel routes, or contrarian opinions precisely because they fit either its predictive model or the user’s declared preferences.

The critical contrast is not between easy and difficult experience, nor between curated and uncurated life. It is between experience selected for its anticipated relevance to us and experience whose presence was not determined by an antecedent account of who we are. Personalization may describe our existing preferences with extraordinary accuracy. But that accuracy remains an incomplete guide for a self that is still becoming. The more our encounters are preselected through a prior account of the self, the more authority who we have been acquires over who we might become.

The answer is neither to abandon personalized technology nor to impose a mandatory cultural canon. Personalization is often genuinely helpful, and no democratic society should require everyone to follow identical itineraries or adopt uniform tastes. The goal is not to manufacture inconvenience, romanticize overtourism, or revive paternalistic gatekeeping.

The goal is to defend spaces where relevance does not govern everything that appears.

Consider the civic institutions that already sustain common fields without demanding uniform responses. A library shelf contains books the reader did not request. A public park brings together strangers who did not select one another. A museum places incompatible centuries within a single line of sight. Public media can present common objects without dictating common conclusions. These spaces are neither uncurated nor immune to bias or exclusion. Their democratic value lies in making substantially common resources available to people who remain free to interpret them differently.

The comparison between Pompeii and Herculaneum reminds us that fame does not make one destination better than its quieter neighbor, nor is the beaten track morally superior to the less traveled path. The vital question is whether a journey leaves room for something neither the traveler nor the system had already identified as a likely fit.

A common world gives us something to share with people unlike ourselves. An encounter not selected to fit us gives us a chance to become unlike ourselves. A culture that personalizes everything risks losing both.

Martina Moneke writes about art, fashion, culture, and politics, drawing on history, philosophy, and science to illuminate ethics, civic responsibility, and the imagination. Her work has appeared in Common Dreams, Countercurrents, Eurasia Review, iEyeNews, Kosmos Journal, LA Progressive, Pressenza, Raw Story, Sri Lanka Guardian, Truthdig, and Znetwork, among others. In 2022, she received the Los Angeles Press Club’s First Place Award for Election Editorials at the 65th Annual Southern California Journalism Awards. She is based in Los Angeles and New York. Follow her on Substack.

This article is licensed by the author under the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License (CC BY-NC-SA 4.0)