Part 3 – Techno-Existential Challenges

The session began with Scott Longfellow, curator of The Solar Biennale, who emphasized the urgent need to diversify the aesthetics of renewable infrastructures. In response to public resistance toward standardized solar panels and wind turbines, he proposed a solar punk vision in which technology and everyday life are integrated through design. “The energy transition also requires a cultural transition,” he said, showing prototypes, photovoltaic materials embedded in architecture, and examples of symbiosis between buildings, algae, and cyanobacteria.

Next was Prodromos Tsiavos from the Onassis Foundation, who presented Plasmata, an exhibition that reimagines digital art in public spaces. Developed over two years with local communities, the project is installed each summer in urban parks in Greece. More than just a show, it serves as a form of placemaking, where technology disappears as an object and becomes a medium to foster encounters, reconfigure space, and leave a trace.

These two projects set the tone for one of the densest discussions of the day: the session on public artificial intelligence infrastructures, moderated by Francesca Bria. Rather than debating hypothetical futures, the block tackled the very real present of AI in Europe—from public supercomputing centers to the tension between regulation, technological sovereignty, and cultural production. The session featured:

• Alexandra Geese, Member of the European Parliament for the Greens, criticized the dominant model of “surveillance capitalism” and advocated for a European alternative based on fundamental rights, sustainable economic models, and open technologies.
• The Barcelona Supercomputing Center showcased its collaborations with artists and cultural institutions, both in creative processes and in work with archives and heritage. “Art functions as an innovation lab,” they said, highlighting projects with María Arnal and the scientific outcomes of these residencies (short video).
• From Germany, representatives from the High Performance Computing Center Stuttgart and the Media Solution Center proposed a co-creation model between science, art, and society. They emphasized the role of art as a gateway to knowledge and called for stronger European collaboration, noting that the first cooperation between two European supercomputing centers was born thanks to an arts program.
• Finally, Bettina Kames, director of the LAS Art Foundation, presented her Sensing Quantum program and a recent exhibition on quantum computing at Kraftwerk Berlin, which attracted 50,000 visitors. “Art can be an entry point to complex topics like quantum physics,” she said. The show included a quantum AI model developed with Google and artist Laure Prouvost.

This session expanded the definition of infrastructure and revealed the challenges and possibilities of an open, public, culturally connected European AI. A contested field, where alliances between artists, technologists, and public policies are still under construction.

Then came one of the most astonishing moments of the day, led by Holly Herndon and Mathew Dryhurst, artists who have spent a decade working at the intersection of artificial intelligence, music, and digital governance. Their practice, which they describe as protocol art, focuses on intervening in systems through the rules that structure them—not just through their outputs. They presented an overview of their most influential projects, from collective training of vocal models at festivals like Sónar to Holly+, an AI trained exclusively on Herndon’s voice (TED talk demo).

Beyond musical creation, they’ve worked to regain autonomy within public AI models. Through experiments like XhairyMutantX, they began by asking if it’s possible to choose how we are represented by AI. They explore ways to insert personal perspectives into algorithmic systems that are typically impersonal and extractive. In parallel, they founded Spawning, an organization promoting alternative frameworks for intellectual property: from the site Have I Been Trained to check if your works are included in massive datasets, to disruptive actions like Kudurru, a system that blocks the data scrapers used to train AI models. This network successfully interrupted image model training worldwide.

In their most recent project, they trained a choral model with 15 choirs across the UK, governed by a data trust that ensures shared control over the generated outputs. With this, they propose a radical vision of AI as public infrastructure, where foundational models can be developed ethically and openly. Their next step: Public Diffusion, an image model trained on 12 million public domain images, as an alternative to platforms dominated by large corporations.

An AI that won’t know who Mickey Mouse is—but will know a lot about mountains.

The penultimate session of the day was dedicated to presenting Eurostack, an ambitious proposal to build European digital sovereignty from the infrastructure up. Backed by the European Parliament, over 200 companies, and governments like Germany and France, the initiative proposes a technological architecture spanning from critical minerals to AI models, through chips, connectivity, sovereign clouds, open-source software, and digital identity. Francesca Bria and Dirma Janse visualize this stack as a regenerative alternative to the dominant extractive model: an interconnected, democratic system that reinforces public services, guarantees digital rights, and places technology at the service of people. More than a technical plan, Eurostack is envisioned as a political, cultural, and ecological project.

The final stretch of the Archipelago Summit was marked by a debate as passionate as it was chaotic. Conceived as a space to imagine possible AI futures, the session ended up polarized into a confrontation between irreconcilable visions: on one side, the defense of sovereign and public European infrastructure against the dominance of big tech platforms; on the other, criticism of academic alarmism that, according to some, prevents concrete solutions from emerging. Benjamin Bratton opened with a provocation:

Let’s not do to AI what Germany did to nuclear energy.

From there, powerful interventions followed, from Evgeny Morozov, Kate Crawford, and Marina Otero, who highlighted the ecological, social, and geopolitical impacts of current models and the urgency of exploring alternatives. However, the tone became so tense that nuances were lost. The transformative potential of AI, the need to regulate it, and the institutional model that should sustain it are, as José Luis emphasized in the closing, arguably one of the most important conversations we can have today.

Conclusion

The Archipelago Summit left me with a renewed sense of responsibility toward the opportunity this complex moment represents. It calls on me to rethink, from the common, how we develop and inhabit this planet.