« Artificial Intelligence and Creation: Stakes and Practices » (free book - UNESCO)

The publication “Artificial Intelligence and Creation: Stakes and Practices” (2025), the result of an unprecedented cooperation between eleven UNESCO Creative Cities from six countries, constitutes an interesting initiative designed to collectively think through the ongoing cultural transformation. The reflection on the intersection of human creation and artificial intelligence is quite open and incorporates controversies, which is rather rare.

A structuring institutional framework

This work is part of a very specific institutional framework: it was initiated by French municipalities belonging to the UNESCO Creative Cities Network, with the support of the French National Commission for UNESCO and the Ministry for Europe and Foreign Affairs, via its Delegation for Local Authorities and Civil Society. This institutional architecture reflects a political awareness that the questions raised by artificial intelligence in the cultural field can be addressed neither by the market alone, nor by States alone, but call for international decentralized cooperation.

The eleven partner cities (Angoulême, Cannes, Casablanca, Enghien-les-Bains, Havana, Lyon, Montreal, Mumbai, Potsdam, Quebec, and Toulouse) represent a great geographical and cultural diversity, and each brings its own sectoral expertise. This distribution of responsibilities according to the specificities of each territory illustrates an approach to international cooperation that trusts collective intelligence rather than centralized expertise. I propose to offer my synthetic point of view here, and it is available for free download.

A methodology of comparison over time

The ambition of this work goes beyond a simple one-off publication. As explained by Maud Boissac, editorial director and director of culture for the City of Cannes, this is the first volume of a ten-year project, entrusted each year to a different country. We shall see if it is done! The idea of this methodological continuity is to compare, over time, the evolution of practices and perceptions regarding AI and creativity. Personally, I think that in 10 years, things will have changed so much that this very approach will be obsolete.

The same six questions were asked of all the experts and artists interviewed, allowing convergences and disagreements to emerge quite clearly:

• Is AI a simple tool or a co-creation?
• To what extent does it affect the authenticity of artistic expression?
• Can it lead to a standardization of works?
• What are the ethical implications of its use?
• How to guarantee innovation face to the risk of algorithmic repetition?
• How to imagine the future of AI in each sector?

This methodology of recurrence implicitly recognizes that we are not facing a one-off event that could be analyzed once and for all, but a continuous transformation whose contours are constantly being redefined. The ten-year project assumes that our answers today will likely be obsolete tomorrow, and that we must therefore institutionalize reflection rather than seek definitive solutions.
This idea of asking the same questions to everyone seemed too simplistic to me at first, but it must be noted that it actually has an interest, which we are about to discover.

Three transversal perspectives to lay the foundations

The book opens with three transversal contributions that set the analytical framework:

• Alexandra Bensamoun, professor of law at Paris-Saclay University, deals with the legal dimension from the perspective of copyright. She reminds us that copyright maintains an original link with technology, as the printing press participated in the awareness of the need for legal protection for creators. Today, the central question is that of value sharing: cultural content used to train AI models has undeniable value, but the authors receive generally no remuneration. The European AI Act of June 2024 imposed a copyright compliance policy and a transparency obligation on providers, but time will tell if this normative incentive is enough to create an ethical market.
• Saida Belaouali, professor of applied ethics and AI ethics at Mohammed First University of Oujda in Morocco, addresses the ethical dimension. She introduces the concept of “interpoiesis” to designate this in-between space, this new locality where human creative expressions dialogue with digital scores. Her analysis highlights that the ethical stake concerns not only technique and aesthetics, but also the manufacture of cultural narratives that redefine our symbolic horizons. Left unguided, this intervention would risk standardizing and impoverishing symbolic architectures.
• Thomas Paris, researcher at the CNRS and associate professor at HEC Paris, proposes an economic reading by evoking the “Promethean moment” of creation. He observes that generative AI abolishes barriers to entry and opens an era of multiplication of supply. Four out of ten people were using AI in February 2025, and Deezer reports that 20,000 tracks designed by generative AI are published every day on its platform. This overabundance risks making access to the visibility necessary for the emergence of a new work or a new voice increasingly costly, paradoxically weakening small players and new creators.

An assumed diversity of viewpoints

What constitutes the democratic value of this work is that it does not seek to produce an artificial consensus. The viewpoints expressed are sometimes diametrically opposed, and that is precisely what makes the reading stimulating. One has the right to have different positions on AI, and this work fully assumes that.

The section dedicated to music illustrates this tension particularly well. On one side, Jean-Michel Jarre asserts that “artists are all thieves”, that he plunders everything he hears, and that from this perspective AI is for him “a modern muse broadening the frontiers of his inspiration,” “a co-creator who dialogues with me, who surprises me and who opens new paths for me”. On the other, Cécile Rap-Veber, CEO of SACEM, holds a radically different discourse. She reminds us that generative AI “has fed on all this culture” without paying the slightest remuneration to creators, and that “it is at the root of the programmed financial impoverishment of creators worldwide with the risk of their extinction.” Sandy Vee, musician and producer, considers that “the notion of co-creator seems exaggerated” and that he “does not see the interest in going on Suno to simply click and receive a track in a few minutes”.

These positions are not reconcilable, and the book does not pretend to reconcile them. That is its interest. Cécile DeLaurentis, an artist and musician member of SACEM, testifies to a practice where AI becomes “an instrument in its own right, a digital entity with which I exchange, I co-compose”, while refusing “to use certain AIs whose practices do not seem ethical to me”. Clément Libes, musician and producer, observes for his part that “any technical assistance that tends to diminish the artisanal involvement of a creative being impacts the authenticity of that gesture and naturally makes it less moving”.

This diversity of positions is found in other sections—cinema, literature, digital arts—with nuances specific to each field. The work does not decide. It exposes. It is up to us to position ourselves.

A partial work that assumes its scope

To be honest, if you are already interested in the subject of artificial intelligence and creation, if you have read texts on the matter, if you have experimented with these tools, you will not learn much that is new by reading this book. The uses described are known, and the questions asked are those that have been circulating in books, articles, conferences, and the media for two years. The metaphor of the mirror, recurring in the book, has been formulated elsewhere. Concerns about standardization, value sharing, and the authenticity of artistic expression are part of the common repertoire of reflection on AI in 2025.

The work is therefore partial, but everything is partial anyway. I find it very good that it assumes its framework; its interest lies in the highlighting of controversy and the diversity of practices, opinions, and societal projects. It is a synthesis of the state of questioning in 2025, seen through the prism of UNESCO Creative Cities and their partners. It does not claim to be exhaustive, nor definitive, nor neutral. It is situated, dated, and inscribed in a precise institutional context. It relates a subject, but above all the state of contemporary viewpoints on a subject.

The importance of the act

What matters to me here is perhaps not so much what we learn as the very act of doing it. The essential thing is to set up instances to think collectively, to confront viewpoints, and to document practices. In this respect, this work produces a form of AI governance. Not regulatory governance, not technical governance, but governance through shared reflection.

When we read this work, we see the diversity of positions, and thanks to this structured diversity, we position ourselves better. There are contributors with whom we agree, and others with whom we disagree. We find our place, illuminating our own choices in resonance with this plurality of voices. This is where the utility of this type of publication lies. It does not tell us what to think; it helps us to think.

This initiative shows that it is possible to produce quality collective reflection on AI, funded by public money, without falling into the traps of technological marketing or the defensive posture of rejection. It also demonstrates that local levels and decentralized cooperation can play a structuring role in the cultural regulation of innovation.

I express the wish that the methodology developed here inspires other similar initiatives. The ongoing cultural transformation needs these spaces for collective reflection, anchored in territories and open to the world, where one can be in disagreement without it being a problem, and knowing that this is how we truly enrich one another.