For one weekend, London's Southbank Centre is turning some of its busiest spaces into a public testing ground for computational creativity. Creative Intelligence, an inaugural event curated by PACT — Planetary Art Culture Technology — brings together artists, digital tools, and artificial intelligence in a programme conceived not as a technical showcase, but as a hands-on experience for visitors.
The initiative arrives in the 75th anniversary year of the London institution, regarded as Europe's largest arts centre. The historical context matters because it clarifies the project's ambition: not to confine generative tools to a laboratory, an industry conference, or a commercial product, but to embed them within the context of a major cultural space open to the public.
At the core of the collaboration with Google Arts & Culture is a question that has become unavoidable for museums, festivals, and institutions: how can one experiment with AI without reducing the artist to a mere input provider and without sidelining people's creative autonomy? The Southbank Centre presents the weekend as a practical, if still necessarily partial, answer to this question. The works do not aim to prove that a model can replace an author; instead, they use interaction to make visible the consequences of the dialogue between human imagination, software, and the public.
Four experiments in the shared space of the Royal Festival Hall
The installations by Google Arts & Culture Lab are hosted in the public foyers of the Royal Festival Hall, specifically in the Clore Ballroom and the Exploration Bazaar. The choice of location is significant: these are spaces of transit and encounter, far removed from the quieter rituals of the traditional exhibition hall. Visitors can participate, play, generate images or sounds, and observe how the machine's response alters the outcome.
Among the experiments showcased is Dreaming with Flamingos, a project by Yinka Ilori that weaves references to West African folklore with Lyria, Google DeepMind’s tool dedicated to music generation. The experience takes the form of a puzzle-based quest: visitors are tasked with restoring lost knowledge to a park and, along the way, unlocking a musical universe tied to the artist’s imagination. In this case, AI is not presented as the sole author of the composition, but rather as part of a narrative and participatory environment crafted by Ilori.
Splash Canvas, created by David Li, brings together Gemini and fluid dynamics. Soft sea creatures become both paintbrushes and commentators on the artwork created in real time by the audience. The creative act, therefore, does not rely on the conventional interface of a chat or graphic design software: it unfolds on a fluid surface, with digital interlocutors actively participating in the image’s creation. It is a meaningful approach because it shifts the focus from the final output to the behavior of the experience itself: it is not merely about generating an image, but about negotiating it through continuous actions, reactions, and transformations.
With Learn Everything, Gemini is used to turn photographs of everyday objects into visual metaphors designed to explain complex topics. The idea is to start from the familiar and recognizable to build a learning path: a common item can become the entry point to a scientific explanation. It represents a potential educational application of generative AI, but it also carries a clear responsibility. While an effective metaphor can aid understanding, it cannot be mistaken for a comprehensive explanation or a verified source, especially when tackling scientific subjects.
Google Arts & Culture also highlights Say What You See among the experiences available over the weekend. In total, four interactive experiments are hosted across public spaces. Their inclusion in the same program allows AI to be viewed not as a single monolith: here it is applied to music, interactive painting, visual learning, and forms of perception and interpretation driven by visitor participation.
A relationship built over time, not a temporary installation
Creative Intelligence does not emerge from a vacuum. The Southbank Centre and Google Arts & Culture have already worked together to celebrate the 50th anniversary of the Hayward Gallery and to digitize Art by Post, a project that delivered artistic activities into homes by mail. The new event builds on that relationship, but shifts both scale and focus: moving from the preservation and digital dissemination of cultural content to the in-person use of systems capable of generating, analyzing, and responding.
This shift clearly illustrates where the cultural sector finds itself today. For years, technology in museums was primarily about high-definition reproduction, online archives, audio guides, projections, or immersive devices. Generative models introduce a different dynamic: the artwork or installation can react to audience input, suggest associations, produce new material, and alter its course during the experience. The promise is a more open experience. The risk is that the novelty factor will overshadow substance, or that the interface will obscure the artistic choices, data, and infrastructure that make the end result possible.
The value of involving artists like Ilori and Li lies precisely in avoiding a generic platform demonstration. A language or music model never enters a cultural space as a neutral medium: it must be framed by creative direction, interaction rules, a distinct vision, and an objective. The audience sees the software, but also encounters the way an author has chosen to limit, steer, or challenge it.
Human creativity remains an open question
The phrase “creative intelligence” may suggest a straightforward alliance between human sensibility and computational power. In practice, the relationship is more complex. Generative systems have become accessible to a broad audience, but their presence in the arts continues to raise questions around attribution, model training, rights, compensation, and transparency. The London program does not claim to resolve them in a single weekend; it does, however, make them harder to ignore by taking these tools off private screens and bringing them into an institution frequented by diverse audiences.
For visitors to the event, the most tangible aspect will likely be the opportunity to interact with the installations. For industry observers, it will be interesting to watch a historic cultural centre use generative interactivity as a mediating language, without necessarily presenting it as the ultimate goal of the artistic experience. It is a useful distinction: AI can expand the formats an institution uses to narrate, educate, and engage, but on its own it determines neither the value of an artwork nor the quality of the relationship with the public.
Creative Intelligence thus offers a European case study on how cultural organisations can host AI tools without relinquishing their critical role. After the weekend, it remains to be seen whether these experiments will achieve continuity within Southbank Centre's programming and, above all, what curatorial, accessibility, and accountability practices they manage to consolidate. For now, London is placing AI in a space where it can be touched, heard, and discussed: a far more useful step than many abstract promises about the future of creativity.
