What Teenagers Can Teach Us About Artificial Intelligence

90 teenagers from 16 cities, 24 hours of dialogue about AI: the ai/teens conference delivers unexpected perspectives. These young people neither fear nor idolize technology. They think.

A conference organized by and for teenagers

On March 15, 2025, an unprecedented experience unfolded. It was a 24-hour global conference, following the sun’s movement from Takasaki (Japan) to Los Angeles (United States), passing through Bengaluru, Yerevan, Kampala, Paris, Berlin, Lisbon, and Buenos Aires. Approximately 90 teenagers and 45 invited adults participated in this event orchestrated by the TUMO Center for Creative Technologies, a nonprofit educational program born in Armenia and now deployed worldwide.

The initiative followed the 2023 ai/education symposium, which had led to an essential question: Why do adults debate among themselves about young people’s future without listening to them? The ai/teens conference therefore reversed the perspective. The teenagers were not passive participants or symbolic witnesses. They were moderators, organizers, and speakers. They questioned researchers, entrepreneurs, and political decision-makers, but also their peers from other continents, comparing how AI shapes school, work, and culture across different contexts.

The time-relay format allowed each city to hold local sessions while connecting to others in real time. From classroom discussions in Montevideo to packed theaters in Mumbai and Kampala, to a debate on Parisian rooftops, the program unfolded in about a dozen languages. More than twenty hours of dialogue, debate, and practical demonstrations circulated through this planetary network, leaving a shared video archive and hundreds of questions for future work.

What immediately strikes is the maturity of the perspectives expressed. Teenagers worldwide share remarkably similar attitudes toward AI, despite their differences in nationality, language, or economic context. Divergences of opinion exist, but they manifest in comparable ways everywhere. As the report notes, “participants had more in common as teenagers than differences related to their nationality, birthplace, or language.”

AI as a “cheat code”: the video game metaphor

A recurring metaphor runs through the discussions: the video game. A participant from Takasaki puts it this way: “Imagine you’re holding a controller and playing a game called life... using AI is like using a cheat code, but that doesn’t mean I’m letting go of the controller.” This image condenses a sophisticated position according to which AI augments capabilities, but human agency remains. The distinction between assistance (acceptable) and substitution (problematic) appears as a fundamental dividing line.

This vision aligns with what I’ve called the concept of “displaced us,” according to which AI is not an external entity facing us, but a version of ourselves that has been displaced and repositioned beside us. It is us, philosophically, culturally, and cognitively, but with a slight sidestep. It’s not a simple imitation; it’s an ontologically shifted us, hence the disturbance it provokes. Teenagers seem to intuitively grasp this hybrid nature, AI as an exoskeleton rather than a prosthesis replacing a missing limb.

The theme of agency traversed all participating cities. In Paris, teenagers warned that becoming too dependent on AI can “break the image of autonomy that develops during adolescence.” Many expressed ambivalence. Yes, AI helps with homework and creative tasks, but this assistance sometimes borders on substitution. A participant from Buenos Aires called on peers to remain “critical creators, not passive consumers.”

A testimony from Yerevan well illustrates the slide toward dependency. “That’s when you understand if you’re too dependent on AI... after every situation, you start thinking, what if I asked AI rather than thinking for myself?” Another participant recounted how the code autocomplete shortcut had become instinctive, almost reflexive. These observations raise a pedagogical question: When does help become dependency? What happens when the scaffolding is removed? The teenagers’ response converges on the fact that AI can assist, polish, visualize, but it cannot bring forth what one doesn’t understand oneself.

This lucidity echoes Jacques Rancière’s vision in “The Ignorant Schoolmaster” (1987), according to which learning is not necessarily a vertical transmission of knowledge but can emerge from a dynamic of empowerment. If machines can instantly provide information and explanations, the role of the educator shifts toward accompaniment and creating environments conducive to autonomous and critical learning. The teenagers at the ai/teens conference seem to have understood this before many adults.

Creativity and authenticity: a redefinition in progress

If AI can compose music, write novels, and make films, what does it still mean to be creative, intelligent, human? This question structured the debates of the “Making It Real” panel. Discussions revolved around a tension: AI as a powerful tool versus AI as a shortcut that dulls hard-won skills. One participant insisted that one must first learn to “write an essay” or “write code” by oneself, then integrate AI into the process, otherwise “someone who hasn’t skipped this fundamental step will obviously surpass you.”

This position aligns with what Michel Serres observed in “Thumbelina” (2012), according to which younger generations develop new ways of thinking and interacting with information, and “cognitive functions transform with and through the medium.” Authenticity, for these teenagers, doesn’t reside in the absence of AI tools but in the intentionality of the person creating. AI can inspire, polish, visualize, but artistic direction and taste remain human.

Several speakers recalled that children learn by copying long before they innovate. The fact that AI learns through imitation therefore cannot be disqualifying in itself. But the difference between real intelligence and the appearance of intelligence lies, according to one teenager, in experience and empathy, qualities that don’t live in data. “There are multiple ways to consider what intelligence is. Intelligence isn’t just data or knowledge in general,” they specified. Another put it more directly: “AI does what you tell it to do,” useful but incapable of the messy leaps that humans make without thinking.

Artist Refik Anadol, a conference guest, expressed a convergent vision. “AI doesn’t do everything for me. I’m the artist. I believe in human-machine collaboration. I don’t believe creativity belongs to AI itself. What I believe is that if I use AI, it can augment my capacity for imagination.” This conception of the tool as creative partner rather than substitute corresponds to what I observe in my filmmaking workshops with artificial intelligence: the quality of dialogue with AI, the choice of models, the formulation of prompts become an integral part of the creative process.

Across all cities, the common thread was agency. Teenagers treat AI neither as an oracle to follow nor as a threat to combat, but as raw material to shape. Being intelligent, in their vocabulary, now includes knowing when to rely on a system and when to rely on oneself and others. This new authenticity embraces tools while preserving intention.

Relationships and friendship: the limits of artificial presence

The “Virtually Friends” panel explored an inevitable question: What makes a friend? For one speaker, it’s emotional safety, being able to talk without being judged. For another, it’s empathy, the feeling that someone not only listens but understands. Can AI fulfill this role? The answers were nuanced. As a teenager from Los Angeles said: “AI is like a smart friend who doesn’t judge me for my questions.” In Beirut, the discussion turned toward AI’s impact on existing relationships. “AI has changed how I communicate with my friends and family. Sometimes it brings us closer, sometimes it creates distance.”

Teenagers don’t reject the idea of a form of companionship with AI, but they set a clear condition: Real friendship involves conflicts, vulnerability, “tough love.” AI can ease loneliness but cannot offer presence—being there in discomfort, without solutions, just with someone. A scenario was evoked: waking from a nightmare at 3 a.m. with no one to talk to. Some admitted they might turn to a chatbot, but none thought it could really help. What they wanted in that moment wasn’t advice or information but presence.

A young participant from Kampala recalled that friendship isn’t just the ideal form of a relationship but also a process that helps people evolve through misunderstandings and lived experiences. AI can offer insights and comfort, but real friendships, with their inherent vulnerability, reciprocity, and occasional conflicts, remain irreplaceable. This distinction echoes what I’ve developed elsewhere: we can maintain rich relationships with machines, just as with animals, without confusing their nature with ours. The enrichment comes precisely from alterity, provided we don’t anthropomorphize it.

Ethical questions also emerged around these interactions. If one confides a secret to AI, is it truly private? One speaker noted that “AI remembers everything.” This permanence can be useful for tracking one’s thoughts but raises a deeper question: What does it mean to confide in something that doesn’t forget? Rather than presenting AI as a substitute for friendship, conversations oriented toward a hybrid vision: using technology to better understand oneself, but reserving the messier and more meaningful work of friendship for humans.

Governance and algorithms: unexpected political pragmatism

Discussions about governance resembled neither a school debate nor theoretical reflection. They were more like a political meeting where those who will live longest with AI’s consequences tried to sketch out the rules before the ink dried. One speaker set the tone: “Teenagers should play a major role in how AI develops. You don’t need to be an expert to decide what’s important and how AI should evolve.” Legitimacy flows from what’s at stake, not from age or degrees.

But the group didn’t abandon itself to naive techno-optimism. A teenager from Amsterdam, who runs AI workshops for younger children, nuanced: “Children don’t know if information is true or not... adults can establish better rules to help children with AI.” Rather than opposing young and old, teenagers kept returning to a more nuanced idea: collaboration that combines youthful curiosity and mature guardrails. They don’t claim exclusive control but a place at the decision-making table.

Regulation occupied an important place. A teenager from Berlin evoked the European legislative momentum. “The EU published a risk assessment... unacceptable risks that must absolutely be banned, like social credit systems or AI surveillance.” Others continued: “We need to create a new organization that regulates AI globally,” imagining a body as global as artificial intelligence itself. But another warned: too much power in the hands of a single entity creates new dangers. “I don’t think a single entity should have complete control of AI... it should be moderated by different groups.” The tension between central oversight and plural governance ran through the exchanges like a fault line.

On algorithms, teenagers acknowledged they shape their identity. Some admitted ending up liking content initially imposed on them by recommendation feeds. But they draw a red line: manipulation. They demand what the report calls “model literacy,” understanding not only what AI says but why it says it. A teenager from Tirana called the algorithm a “builder because it builds paths for us,” but also a “manipulator” when misused. The demand is clear: transparent rules, visible to all, adaptable according to cultures, and above all no hidden surveillance.

Education: a transformation already underway

From Kampala to Los Angeles, teenagers expressed the same observation: they’re not waiting for adults to decide whether AI has a place in school. They’re already using it, questioning it, and above all expecting it to be treated as seriously as any other fundamental subject. A speaker from Takasaki put it clearly: “You should learn AI like math, English, or Japanese.” Mastery begins with basic literacy, not alarms about plagiarism. In Tirana, students emphasized the importance of understanding how AI works, how it impacts society, and what its ethical implications are.

A spirit of pragmatism manifested in how teenagers adapt AI to their own interests. In Amsterdam, one participant joyfully asked a chatbot for “Dutch grammar exercises themed around Dragon Ball Z,” because a touch of anime makes practice bearable. In Buenos Aires, another transformed an entire textbook into a custom GPT “so everyone can learn in the way that suits them best” and, crucially, “so I can share it with my friends.” Personalization isn’t a marketing slogan; it’s a workaround, an antidote to uniform courses.

A teenager from Berlin recalled the overall situation: “The real problem isn’t that AI is changing education. The problem is we’re still learning like 100 years ago.” This frustration ran through many panels. Teenagers don’t fear automation so much as obsolescence. They’re impatient to join classrooms that reward the best questions rather than the fastest answers. This perspective aligns with what John Dewey called in “Democracy and Education” (1916) “progressive education,” capable of adapting to individual needs while pursuing collective purposes.

Sal Khan, founder of Khan Academy and conference guest, shared that technology, including AI, can bridge access gaps, but its primary role in traditional classrooms shouldn’t be to replace human interaction. He sees AI as a liberating force, allowing teachers to move beyond lectures and foster more interactive, human-centered learning experiences. Khan also emphasized that essential skills will include solid academic foundations, but especially entrepreneurial capacity—not just creating businesses, but using existing resources and new tools to create value in unprecedented ways.

What emerges from these scattered remarks across the world is a coherent agenda. Teenagers ask educators to shift from surveillance to accompaniment, from fear of cheating to a culture of judgment. The challenge is no longer so much transmitting knowledge as developing meta-cognitive skills: learning to learn, to question, to validate, to create meaning. If adults don’t act, teenagers will continue to cobble together their own solutions: avatars for shy students in Takasaki, improvised campus chatbots in Amsterdam, community GPTs in Buenos Aires.

Seven lessons for rethinking our relationship with AI

The report synthesizes seven major themes emerging from the discussions:

1. The first, “a cheat code for agency,” reveals a sophisticated understanding of technology as an augmentation exoskeleton rather than a crutch. The goal isn’t to avoid AI assistance but to maintain intentionality and agency while using it.
2. The second, “a new authenticity,” proposes a framework for creativity that embraces tools while preserving the intentionality of the person creating.
3. The third lesson, “keeping friendship real,” establishes a clear test: it’s not enough to offer advice and support; conflict and tough love are inherent to real friendships.
4. The fourth, “surprising pragmatism,” shows that teenagers approach AI governance with striking political maturity, advocating for collaboration that combines youthful curiosity and mature guardrails. They don’t reject a key role for adults but demand meaningful participation in decisions that will shape their future.
5. The fifth lesson, “living with the algorithm,” recognizes that algorithms don’t just predict behaviors; they shape them. Teenagers seem open to the idea that their identity might be co-constructed, but they draw a firm line at manipulation.
6. The sixth, “self-emancipation in education,” perhaps constitutes the most subversive observation: instead of waiting for permission to integrate AI into learning, they “hack” their own solutions. Pedagogical transformation will be carried by learners, not by schools.
7. The seventh lesson, “denouncing false dilemmas,” perhaps best sums up this generation’s stance. Teenagers reject the polarized discourse that seems to dominate adult discussions about AI. They refuse to choose between human creativity and AI creativity, dependency and autonomy, freedom and regulation. They favor agency, transparency, and intentional collaboration. They don’t need to frame the discussion with forced binary choices. Neither utopia nor catastrophism, but discernment.

This generation we worry is most influenced by AI turns out to be the most thoughtful when it comes to thinking critically about it. This paradox of youthful sophistication is striking when watching teenagers take charge of the conversation about AI. They perceive the anxiety of the adults around them and ask to be trusted to do the right thing. Their careful analysis of AI’s complexities is accompanied by lucid optimism that contrasts with the panic dominating much public discourse. As the report notes, “parents, teachers, and other adults who seem to talk to them and talk above them about AI need to stop and listen.”

What this voice teaches us

The ai/teens conference offers a valuable resource for anyone working on education and AI issues. It brings a complementary perspective to that of experts—the perspective of those primarily concerned. These teenagers don’t see AI as a singularity but as part of a digital continuum with the Internet and social media. Many of the social, ethical, and cognitive consequences seem familiar to them. They perceive AI as adding a new dimension rather than creating a separate set of problems.

Throughout the discussions, teenagers systematically consider AI as a public good. They’re not intimidated by the large companies competing to commercialize the technology, nor by governments eager to regulate and control it. They speak of AI as something that belongs to them. This sense of ownership seems self-fulfilling: they don’t see themselves as passive consumers of AI but as actors in its development. This mindset authorizes them to set boundaries, demand transparency, and imagine a future where AI serves their values.

The report formulates concrete recommendations:

• Create “curriculum design labs” co-led by young people.
• Pilot “AI reflection classes” where every interaction is questioned.
• Establish youth advisory councils with real decision-making power.
• Launch “teenage juries” to evaluate emerging AI tools.
• Guarantee real access (connectivity, devices, electricity).
• Fund AI projects led by young people, particularly in the Global South.
• Respect local cultures and languages.

These recommendations share a common thread: they ask adults to stop talking about teenagers and start building with them. Our responsibility isn’t to deny the reality of ubiquitous AI or to suffer it, but to transform it into an educational opportunity. As the report summarizes, “teenagers must be active partners in how AI is developed and used. This means building approaches that make room for shared learning, co-creation, and ongoing involvement, so that their perspectives and experiences help shape the world they will inherit.”