UNESCO’s recent programme on creativity and artificial intelligence brings a deeper question into view: when producing cultural material becomes easier, what determines whether human creators, local voices and diverse traditions can still reach audiences and sustain a living?
Imagine an independent musician preparing a new release. Assistance with translation, visual experiments or an early arrangement could make an ambitious project more manageable. Yet none of those conveniences answers the questions that follow: will listeners encounter the work, understand its origins and support the person who made it? This is an illustrative situation, not a documented case. It captures a distinction that is becoming central to the social meaning of artificial intelligence. Making cultural material and maintaining a cultural life are related activities, but they are not interchangeable. An increase in the first does not automatically secure the conditions of the second.
The UNESCO programme for its international congress of 27 October 2026 places that distinction on the policy agenda. Organised in Paris within the framework of the Convention on the Protection and Promotion of the Diversity of Cultural Expressions, the event was designed around human creation, rights-respecting AI and the discoverability of diverse content. The published programme establishes the subjects and intended dialogue; it is not evidence that participants agreed on a regulatory settlement or that the difficulties discussed have been resolved.
Its importance is the breadth of the question. Public discussion often asks whether a machine-generated picture or song deserves to be called creative. That philosophical argument matters, but it can obscure more immediate decisions about access, recognition and livelihoods. A society might find synthetic material entertaining while still wanting people to develop distinctive artistic practices, communities to interpret their own traditions and audiences to encounter more than the most commercially predictable forms. These goals require attention to the relationships surrounding creation, as well as to the capabilities of a model.
A larger supply is not the same as a more diverse culture
Generative tools can lower some of the practical barriers between an idea and a presentable experiment. Someone without a production team may explore a visual direction, rehearse alternative wording or test how an explanation might sound in another language. Used thoughtfully, such assistance could expand participation. But cultural diversity is not simply the number of files available. Ten thousand variations on a familiar aesthetic do not necessarily provide more meaningful choice than a smaller collection of works rooted in different experiences. Volume measures abundance; diversity also concerns which experiences become visible, intelligible and valued.
The congress’s concept note explicitly connects AI with cultural creation, distribution, participation and the economic models sustaining creative sectors. It identifies remuneration, transparency, recommendation bias and digital inequality as issues for discussion. These are stated concerns and objectives, rather than quantified findings about how much diversity AI has already removed. Their connection is nevertheless analytically useful: an intervention that helps someone produce more work can still leave untouched the conditions that prevent that work from finding an audience.
Consider an illustrative regional publisher able to translate descriptions of its books more quickly. The benefit may be genuine, but international sales still require trustworthy translation, editorial judgement, distribution and reader attention. A generative service can help with one stage without solving the others. The same applies to a filmmaker developing a treatment or a community organisation documenting a performance. The socially relevant measure is not how many outputs the tool supplies. It is whether people acquire a stronger capacity to communicate something that matters to them and to retain a meaningful role in the process.
Discoverability becomes part of cultural freedom
There is a difference between a work being available and a person being likely to encounter it. A recording can exist on a platform without appearing in a search result, recommendation or curated collection. When a listener asks an assistant for something to hear, the assistant’s interpretation of the request becomes part of the route between creator and audience. This introduces a practical question about mediation: which descriptions, categories and sources help a system recognise the relevance of an unfamiliar work? It also raises the possibility that what is easiest for software to classify becomes easier for people to discover.
For this reason, discoverability cannot be reduced to telling artists to improve their promotional skills. A creator can supply accurate information and still depend on decisions made elsewhere about ranking, presentation and access. An audience member may also lack the vocabulary needed to request a cultural form they have never encountered. Serendipity, knowledgeable curation and translation can matter precisely because they introduce something beyond an existing preference. If recommendation is judged only by the likelihood of an immediate response, institutions should ask what kinds of cultural encounter that measure overlooks. This is a design question, not a claim that every platform necessarily narrows experience.
A more diverse environment would make the route to an audience open to examination. Cultural organisations could investigate whether their catalogues are represented accurately, whether searches in different languages retrieve relevant material and whether smaller contributors can correct mistaken descriptions. Public institutions could support cataloguing and translation alongside production grants. None of these measures guarantees attention, nor should cultural policy promise every work an audience. Their purpose would be to reduce avoidable barriers and make the conditions of visibility less dependent on resources that only established organisations possess.
Recognition and permission are different relationships
A cultural work can acquire value through the association between an expression and its maker. A listener may care about who sings, what experience informs the song or how a performance belongs to a place. Automated generation complicates that relationship when audiences cannot tell how material was produced or what human contribution it contains. Clear explanation can help, but transparency has to be specific enough to be useful. Saying that AI was involved leaves open whether it corrected a recording, generated a substantial passage or supplied most of the final performance.
Permission raises a separate issue. Being credited does not necessarily mean that a creator has agreed to a use, and an agreement about one activity does not automatically settle another. The UNESCO concept note identifies consent and the use of copyright-protected works across the AI lifecycle as subjects for discussion. That agenda should not be mistaken for a statement of uniform law across countries. At the practical level, however, the distinction is easy to understand: people need to know what a proposed arrangement allows, how its scope can be checked and what happens when their work is used outside it.
The imbalance between a small creator and a large service makes those questions particularly consequential. An individual may lack the time or expertise to negotiate an elaborate contract or investigate disputed use. Collective organisations, accessible advice and understandable terms could therefore matter as much as a new technical control. At the same time, preserving space for experimentation and legitimate reuse remains important to cultural life. The challenge is to build workable relationships around contribution and permission, rather than assume either that every restriction protects creativity or that every removal of friction benefits the people who create.
Creative livelihoods depend on an ecosystem
A related account from the World Creativity Organization, published on 27 October, reports on the World Conference on Creative Economy in Jakarta earlier that month. It describes an emphasis on the connections between skills, finance, market access, partnerships and digital tools. As a participant organisation’s account, it offers a perspective on the discussions, not an independent evaluation of their effects. Its useful contribution here is to place AI within a wider set of conditions needed to turn creative potential into sustained opportunity.
That perspective helps explain why a cheaper production process may have very different consequences for different people. An established studio could use assistance to explore more ideas while retaining paid specialists to select and develop them. A freelancer could encounter clients expecting a lower price because they believe the work now takes little effort. A newcomer might gain the ability to demonstrate a concept yet still lack the relationships needed to receive a commission. These are plausible scenarios, not reported outcomes from the conference. They show why the effect on a livelihood depends on bargaining, demand and recognition alongside technical efficiency.
Institutions supporting creative activity should therefore look beyond training people to operate the latest tool. An artist who learns a useful workflow may still need affordable space, reliable payment, professional networks and access to audiences. A community archive may need governance and maintenance funding more urgently than another demonstration project. Connecting these elements is difficult because they sit within different budgets and organisations. Yet treating AI as a complete development strategy risks supporting activity that cannot survive once an initial grant or promotional moment has passed.
Communities need influence over how they are represented
The cultural stakes become especially visible when the material involved is linked to collective identity. A tradition may include stories, techniques or forms of expression whose meaning depends on who shares them, with whom and in what circumstances. Turning those expressions into accessible digital material can support learning and exchange. It can also strip away the context that distinguishes participation from imitation. A model’s ability to reproduce a recognisable surface does not establish that the resulting work carries the same significance or that its presentation respects the people associated with it.
There is no reason to assume that communities will all want the same response. Some may welcome broad experimentation; others may distinguish between public knowledge and material shared under particular responsibilities. Even within a community, views may differ across generations and artistic practices. Meaningful participation therefore requires more than obtaining a single endorsement and treating it as permanent permission. It calls for arrangements through which people can explain concerns, revise descriptions and take part in deciding what a project is for. Those arrangements should recognise disagreement rather than invent a cultural authority that nobody actually holds.
This is also where the language of preservation needs care. A living culture is not simply an archive of patterns to keep unchanged. People reinterpret inherited forms, combine influences and disagree about what counts as a valuable continuation. The role of responsible AI adoption would be to strengthen their capacity to make those choices. Success could involve a community using digital assistance on its own terms, or deciding that a particular use is inappropriate. Either outcome may represent greater agency than a project that produces impressive material while leaving the people concerned without influence.
The next measure of progress is cultural agency
The Paris programme and the wider creative-economy discussions provide an agenda, not a completed demonstration of social benefit. Their value will depend on subsequent decisions and the evidence gathered about them. Evaluation could examine whether smaller creators gain sustainable commissions, whether catalogues become easier to explore across languages and whether participants can understand and influence the terms of a project. It should also ask who is absent. A service may look inclusive among its users while excluding people who lack connectivity, accessible interfaces or confidence that their interests will be respected.
AI can become a useful instrument in cultural life without being the author of the goals that life should serve. The central opportunity is to expand human capacities for expression, exchange and interpretation. The central risk is to confuse an increase in synthetic production with an increase in those capacities. What matters is whether people can make and share work that reflects their experiences, encounter unfamiliar perspectives and sustain the relationships through which creative knowledge develops. As cultural institutions respond to the new tools, that human agency offers a more demanding and more revealing measure of progress than the quantity of material a machine can generate.
Image: Small Bedroom Recording Studio by Dejan Krsmanovic, CC BY 2.0, via Wikimedia Commons. Illustrative photograph of a creative workspace, not a UNESCO event or a demonstrated AI deployment.