Cultivating empathy and controversy: my proposals for human governance of ai in cultural institutions

After twenty years of actively supporting cultural institutions through their digital transformations, I share here my convictions and methods for keeping artificial intelligence in service of the human community rather than the reverse. Intervention given at the launch of the Observatory on AI in Arts and Media (ObsIA).

The ObsIA

On October 17, 2025, the Arts & Media Department at Sorbonne Nouvelle launched the Observatory on AI in Arts and Media (ObsIA) during a Franco-Quebec study day. Initiated by Laurence Allard (IRCAV), Kira Kitsopanidou (IRCAV) and Alberto Romele (IRMÉCCEN), this observatory aims to develop critical approaches to artificial intelligence, moving beyond consensus discourse to question AI’s present: its social and environmental stakes, emerging resistances, and situated practices.

The day brought together French and Quebec researchers, cultural professionals, and artists. I presented an intervention on “Active and Engaged Institutional Monitoring of AI and Culture,” alongside notably Filmatters (understanding AI through cinema) and Ecoprod (environmental impact of audiovisual AI). The ObsIA positions itself as a space for transdisciplinary and transatlantic reflection, privileging the analysis of contemporary practices over abstract speculation.

My Journey: From Mobile Innovation to Artificial Intelligence

I began exploring technological disruptions in culture in 2005, when cameras had just appeared in mobile phones. With the Forum des images, we created the Pocket Film festival, the world’s first festival dedicated to creation with mobile phones. At the time, many judged this idea absurd. Today, the vast majority of visual content is produced with smartphones.

This foundational experience taught me a lesson I’ve systematically applied ever since: never say “it’s bad” or “it’s good” in the face of innovation, but always ask “how can we do best with what’s happening to us.” This pragmatic stance guides my work today on artificial intelligence with various cultural institutions, from the Centre Pompidou to local authorities.

I facilitate territorial action-research projects, local action-training programs, an AI monitoring committee at the Forum des images currently, I support creative experiments and organizations in their transformations, I train teams... In all these contexts, I observe the same thing: the institutions that succeed in their transformation are those that accept complexity, assume their contradictions, and above all, cultivate two fundamental qualities: the capacity for productive disagreement and empathy between people.

My Method: Intense Preparation to Liberate Speech

When I facilitate a working group, particularly on AI, I always begin with intensive individual preparatory work with participants beforehand, individually or in small groups (videoconferencing is ideal for this), not to format their thinking, but to help them clarify what they really want to say, and then be capable of synthesizing it in a short, dense form. This preparation is not a constraint but a liberation: when you know precisely what you want to express, you dare more to confront it with others.

I impose strict speaking times (seven minutes maximum, for example) with a clearly visible clock on the table, which acts as a collective rule of the game, which we can play with. This temporal constraint, which some might find rigid, actually produces a paradoxical effect: it makes everyone accountable and intensifies exchanges. No one has time to get lost in generalities, everyone goes to the essential. But beware: I never cut anyone off. The clock is a collective indicator, not a guillotine. If someone goes over, it’s their choice, their responsibility toward the group, I leave people free.

I systematically practice real-time mind mapping, projected on a large screen. While participants speak, I map their ideas, which we see being written on the screen, creating a visual architecture of the discussion. This technique, which I’ve refined over the years, serves several functions: it objectifies exchanges, aids our memorization, allows remote participants to follow. But above all, it makes visible the collective thought being constructed, creating unexpected bridges between ideas, and sparking new ideas that might not have emerged otherwise.

My Primary Conviction: Disagreement as the Engine of Collective Intelligence

I categorically refuse the search for soft consensus. In my facilitations, I actively encourage the expression of disagreements, tensions, contradictions. We are in a democracy, and controversy is not only legitimate but necessary. Facing AI, pretending there exists a single position, a “right” way to do things, is intellectual imposture.

I carefully document these disagreements. In my syntheses, I never seek to erase them but on the contrary to signify their value. A report presenting three well-argued contradictory positions is infinitely better than a fuzzy consensus where no one truly recognizes themselves. This approach, which I systematically defend to sponsors, produces richer, more honest, more useful documents.

Concretely, I sometimes organize “structured disagreement” sessions where I explicitly ask participants to formulate their objections to others’ positions. These moments, far from being conflictual, often become the most creative: it’s in friction that new ideas emerge. Assumed disagreement is infinitely more productive than polite acquiescence that hides underground resistances.

This culture of controversy seems particularly necessary to me when facing AI. Extreme positions, whether blissfully technophile or apocalyptically technophobic for example, prevent us from thinking in nuance. I encourage instead a form of permanent intellectual discomfort, where everyone accepts questioning their certainties.

My Deep Conviction: Empathy as Creative Resistance

But controversy alone is not enough. It must articulate with what I consider the key to everything: authentic empathy between participants. This empathy is not a supplement of soul, a surface politeness. It’s the very condition of productive collective thinking.

I cultivate this empathy through concrete gestures. I remember this meeting where, faced with a dysfunctional room configuration, I found myself on all fours on the huge conference table to move a 360 camera. This moment of assumed vulnerability, a consultant in a suit on all fours in front of institution directors, transformed the atmosphere. By daring, I created a space where everyone could dare in turn. Some participants told me afterward that they understood my method through this gesture.

I systematically organize informal times, long breaks where participants can exchange freely. I ensure that everyone is listened to, truly listened to, not just heard. When someone expresses a concern, even clumsily, I reformulate it to ensure it’s understood by all. This attention to the human dimension of exchanges is not a waste of time, it’s an investment in the quality of collective reflection.

The empathy I defend is not soft compassion but a form of active attention to the other. It includes the capacity to tell someone you disagree with them while deeply respecting their position. It presupposes accepting to be transformed by the encounter, not camping on one’s initial positions.

My Approach to Paradoxes: Assuming Contradictions

I refuse binary positions on AI. Yes, I use an AI to synthesize meetings about AI, and I see there not a contradiction but coherence: experimenting with what we’re talking about. Yes, some cultural institutions accept funding from private multinationals, while developing critical reflection on these actors. These paradoxes are not weaknesses but realities to assume.

I prefer a thousand times an institution that says “we are funded by X and we nevertheless develop critical reflection on X” to an institution that claims impossible purity. We are all compromised somewhere: we use smartphones, we have social media accounts, we benefit from digital conveniences. Honesty consists in recognizing these compromises and reflecting from them, not despite them.

In my interventions, I encourage institutions to make their dilemmas explicit rather than masking them. When a museum uses AI to personalize visitor paths while questioning surveillance of audiences, I propose making this dilemma visible, even making it an object of mediation with visitors.

My Concrete Proposals for Institutions

Based on these convictions, I recommend several concrete actions to cultural institutions:

• Create spaces of assumed controversy. Regularly organize internal debates where divergent positions are not only tolerated but encouraged. Document these disagreements, make them public. Show that the institution thinks, doubts, searches, rather than claiming to hold certainties.
• Invest in active empathy training. Train teams not only in AI technical tools but in relational skills that allow maintaining the human link. Learn to truly listen, to reformulate, to welcome discomfort, to create trust.
• Experiment with transparency. Test AI uses but publicly document these experiments: what works, what fails, what worries. Create accessible logbooks where the institution shares its trial and error.
• Ritualize moments of vulnerability. Institute moments where leaders, experts, knowers accept not knowing, being wrong, learning. These moments of shared vulnerability create the conditions for genuine collective intelligence.
• Maintain non-automated spaces. Clearly identify what, in the institution’s functioning, must remain irreducibly human: certain decisions, certain creative processes, certain moments of encounter. Fiercely protect these spaces against the temptation of technical efficiency.

My Plea: Antifragility Through Human Connection

Facing AI, I advocate for what I call institutional antifragility based on the quality of human connection. The institutions that will survive and prosper are not those that will have the best technical tools, but those that will have managed to maintain, enrich, and deepen in the field of the sensible the links between the people who compose them.

This conviction is rooted in my experience. I’ve seen technologically perfect projects fail for lack of trust between actors. I’ve seen makeshift experiments succeed brilliantly because the team was united. Technology is only an amplifier: it amplifies our capacities as much as our dysfunctions.

History teaches us that pure resistances are doomed to failure. Music streaming swept away an industry that had refused to think about digital transformation. Today, refusing to think about AI condemns us to suffer it. But thinking about AI doesn’t mean submitting to it. It means developing a collective intelligence capable of negotiating with it, domesticating it, putting it in service of our human projects.

My Vision: Keeping AI in Service of the Human Community

For AI to remain in service of the human community and not the reverse, I’m convinced we must simultaneously cultivate two apparently contradictory qualities: the capacity for disagreement and the capacity for empathy. Disagreement without empathy becomes sterile violence. Empathy without disagreement becomes soft consensus. But their articulation produces this particular form of collective intelligence we urgently need.

Concretely, this means creating spaces where one can say “I disagree with you” while maintaining the link, where one can express fears without being judged, where one can experiment without being condemned. These spaces are not decreed, they are patiently built, through gestures, attentions, rituals.

I propose seeing cultural institutions as laboratories for this new form of governance: neither technocratic nor technophobic, but profoundly human in its capacity to welcome complexity, to maintain fertile tensions, to transform contradictions into creativity.

The stake is not to resist AI but to resist with it dehumanization. This resistance passes through the obstinate maintenance of what makes our humanity: our capacity to disagree, our capacity to bond, our capacity to create the unexpected. As long as we cultivate these capacities, AI will remain a tool. If we abandon them, we will become AI’s tools.

My commitment, through my interventions, my trainings, my support work, aims precisely at this: helping cultural institutions remain profoundly human places, where controversy and empathy nourish each other mutually to produce this collective intelligence our era desperately needs. Because fundamentally, the question is not “what can AI do?” but “what do we want to do together, with or without AI?”. And this question can only be answered by a human community bound by empathy and enriched by its disagreements.