The trick of Refik Anadol’s Machine Hallucinations, a three-day public art installation at The Shed in New York City, is to transform the processing of data into surreal hypnosis. The immersive audiovisual exhibit towers over a cavernous 17,000 sq ft gallery in Hudson Yards, an outer ring of screens features a shimmering and chameleonic display of what looks like pixelated sand. But each square is a narrative of data: a familiar image – tree, building, lamppost, over 130m publicly available images of New York City searched and collected by Anadol and his team’s algorithms – morphed into a single-colored square and then silenced by a single question: what would you do if you owned your data?
The free exhibit, part of a $250m project to shift data ownership from private mega-corporations to individual users called Project Liberty, makes a tactile, sensory, emotional argument for data dignity and decentralization of internet power – concepts often so bogged down in technicality, abstraction and vagueness as to be inaccessible.
The overarching aim of Project Liberty is to imagine an internet future not governed by tech CEOs, the forfeit of your data for participation, surveillance capitalism and the whims of social media companies aiming for infinite scale. The project encompasses high-level manifestos, a speaker series, and a new open-source protocol, the Decentralized Social Networking Protocol, to serve as infrastructure for a more egalitarian internet.
Machine Hallucinations, a piece of public art premiering to a public largely starved of communal art experiences, offers somewhat of an entree to the philosophical argument. The installation, an expansion of Anadol’s installation at Chelsea Market’s Artechouse in 2019, takes the shadowy, often intimidatingly complex concepts of data ownership and digital footprints and “makes it more legible, more accessible, so there is that ability to engage people in a conversation that is more universal, that it’s not just a bunch of technologists talking to each other,” Frank McCourt, CEO of McCourt Global Inc and the project’s founder, told the Guardian.
The argument is subtle coaxing ringed with overt messaging. Anadol’s algorithms mimic imaginative play with hordes of real images in a deliberately trance-inducing experience, like if you combined the hypnotic effect of a lava lamp with the vertigo of peering over a waterfall, while smaller screens around the room offer stark mantras for reframing data ownership – “your data is BEAUTIFUL”, “your data is POWERFUL” – reminiscent of body positivity slogans. QR codes lead to the project’s website of programming, speakers at this week’s Unfinished Live conference, and mission statement to “create a new civic architecture for the digital world”. The core of the installation contains two screens, one representing the current data landscape (no personal control) facing another to indicate the imagined future (alight and shimmering at the gesture of your hand) that unveils a message with the sweep of your hand: “YOU CONTROL YOUR DATA.”
Anadol, born in Istanbul and now based in Los Angeles, sees our vast data footprints – photos, geotags, transactions – as “a form of memory”, he told the Guardian. “Can we use our collective memories to make collective dreams, to incite collective consciousness?”
Machine Hallucinations, made with Anadol’s team of 14 studio assistants, attempts to put shape to the input process of an algorithm – a playful, whirring visualization of artificial intelligence taking in and learning from images. A custom algorithm trawled the internet for images of New York on social media, search engines, digital maps – all available on public domains, owing to Anadol’s privacy concerns (and to hammer home the point that this enormous amount of information is available without breaching passwords). Another algorithm then cleared the images of people – no passersby, no pedestrians, no faces or bodies. “We let AI learn from that New York, the urban memories of New York, from different seasons, different times, different perspectives,” said Anadol.
“What will happen if AI has the ethics of not seeing a human, but New York as a memory, to dream and hallucinate?”
The intent, he said, was to offer imaginative space to the often impenetrable and droll discussions of tech infrastructure and data. Tech companies have skated by on the numbing effect of onerous explanations, and the assumption of complexity – the 100-page user agreements, the fine print that’s somehow understated and overbearing at once, the automatic location services, the cookies you have to accept.
“We are all tracked by the systems, by hardware and software – what we eat, what we say, what we watch, where we go, what we read, these are all defined by algorithms around us. So of course privacy is at stake,” said Anadol. Machine Hallucinations doesn’t so much argue against the current digital landscape as push toward a more capacious understanding of a public, open-sourced future. “What else can we learn better, remember better, dream better?” said Anadol.
“We need a new civic architecture that responds to that and accommodates that [digital] world – a tech architecture that creates value for society, not one that extracts value from society,” said McCourt.
Both described Project Liberty, and Machine Hallucinations, as an “optimistic” project, an apt word for both the belief in collective action to wrest power from the consolidators of the internet’s wild west and the feeling of being swallowed by the chasmic music and mesmerizing colors of the installation’s cavernous space.
“We are, as humans, the meaning of that data, those numbers,” said Anadol. “It’s in our hands to train these machines with our own dreams.” Such dreams could be an open-source infrastructure for data ownership, the aesthetic potential of AI, or just a return to a public art space in which the familiar – echoes of an old screensaver, images of the city you live in – are cast anew.