The Serpentine Bets on Research, Not Results

There is a particular kind of art funding that asks for nothing to be finished. No exhibition at the end, no object to insure, no press release announcing a completed work. It is rare, it makes trustees nervous, and this week the Serpentine put £40,000 behind it.

On Tuesday the gallery named the first four fellows of its Future Art Ecosystems R&D Fellowship: Alice Bucknell, Reina Mun, Alfredo Salazar-Caro and Auriea Harvey. Each receives £10,000 and six months of mentorship, running from this month to March 2027, structured as a low-residency programme with three weekend gatherings in London and a weekly online rhythm in between. Nearly a thousand people applied. The theme is “Art x Convergence”, and what the fellows are being paid to do is think about artificial intelligence — not illustrate it, not decorate it, but interrogate what it is doing to the systems the rest of us live inside.

What the fellows are actually looking at

The four proposals are more concrete than the framing suggests, which is a relief. Alice Bucknell, an artist and writer whose work has appeared at the V&A and the Centre Pompidou, is examining “world models” — the AI systems that simulate an environment and predict how it will change, used for everything from training robots to long-range climate forecasting. Her interest is in what those models leave out, on the grounds that omissions reveal values. The research is heading towards a multiplayer video game planned for late 2027.

Auriea Harvey, a digital artist and sculptor of thirty years’ standing who now works in Rome, is asking what happens to the human figure when it exists permanently as a file — authored in one place, fabricated somewhere else, resolved as physical or digital depending on where it lands. Alfredo Salazar-Caro, co-creator of the VR institution DiMoDA, is working on where computational design and robotic earth construction meet Indigenous and vernacular building traditions. Reina Mun, a doctoral researcher at Harvard, is taking the technology industry’s central promise — that these tools enhance human capability — and treating it as an open question rather than a given.

That last one is the most quietly radical thing in the programme. Serpentine has assembled a jury that includes Google DeepMind’s chief AI readiness officer and the chief executive of Snap, alongside artists Refik Anadol, Cao Fei and Holly Herndon and the economist Mariana Mazzucato, and then handed money to someone whose stated project is to test whether the enhancement claim holds up.

Why a gallery is doing this at all

Serpentine has been running its Arts Technologies programme since well before the current AI moment, and Future Art Ecosystems has existed since 2019 as a research strand concerned less with artworks than with infrastructure — the contracts, funding structures and institutional habits that determine whether ambitious technological art can be made at all. The argument, roughly, is that critical commentary on AI is abundant while the practical scaffolding for artists working with it is thin. A fellowship that pays for process rather than product is an attempt to build some of that scaffolding.

The money comes from Glass Castle, a new philanthropic body chaired by Suneil Setiya, co-founder of the investment firm Quadrature Capital. It is worth naming plainly: research into how AI reshapes culture, funded by finance, adjudicated in part by the companies building the technology. None of that invalidates the work, and the fellows chosen do not look like people who will produce flattering conclusions. But the shape of the arrangement is now the normal shape of ambitious art funding in Britain, and readers of this journal have watched it recur — in acquisition budgets, in prize money, in who ends up on boards.

What it means further down the food chain

For an independent gallery in Bristol or Leeds, a £10,000 research fellowship in Kensington may look like someone else’s conversation. It is not, quite. Fellowships like this one set the terms on which a whole category of work becomes exhibitable. Artists working with simulation, agentic systems, XR and computational fabrication have historically struggled to find venues, partly because nobody had solved the boring questions — how you install it, who owns the output, what happens when the software stops being supported, how you price something that exists as a file. Serpentine’s fellows are being asked to document their process publicly and share it as it develops, which means those answers should surface in the open rather than staying inside one institution’s files.

There is also a useful reframing here for anyone programming a small space. The fellowship’s own material makes a distinction worth borrowing: convergence, in the technological sense, tends to narrow towards a predetermined goal, while art branches outwards and moves deliberately through instability. Optimisation forecloses; art diverges. That is a decent description of what a good independent gallery does anyway, and a decent argument for why the smallest rooms in the country remain the places where the most genuinely unresolved work gets shown.

The first London gathering happens this month. The public sharings come later, and they are the part worth watching.