I am standing in the central hallway of the National Gallery as the Christmas markets go up outside in Trafalgar Square. The Gallery’s director, Gabriele Finaldi, is unveiling its latest collaboration with Google Arts & Culture (GAC), the non-profit arm of the world’s fourth biggest company. He’s here with GAC’s founder, Amit Sood, and the two beam with pride as they introduce a crowd of journalists, patrons and trustees to the website’s new features, which include online exhibits on artists and famous paintings and an AI tool composing soundtracks to the collection’s most celebrated artworks. Technically speaking, it’s impressive.
GAC has been a partner of the National Gallery for more than a decade. The website today counts 50 million users and is also available as an app (one million downloads, including my own, dating back to 2019 when GAC launched a feature to determine which seventeenth century portrait you most resembled based on a selfie submission). Over 13 years, GAC has democratised access without precedent to some of the world’s most prestigious collections. Anyone with an internet connection can browse the artworks on display at the Tate Galleries, MoMA and the Musée d’Orsay, zooming into details and studying the minutiae of artefacts that date as far back as the 50th century BC (in the case of the Leang Karampuang cave painting in Indonesia).
The National Gallery is currently hosting a landmark exhibition on Van Gogh, arguably the greatest colourist of all time; now, his and all works in the Gallery can be viewed through the prism of colour, thanks to a new feature on GAC that classifies paintings by palette. Select gold, for instance, and the boundaries of time and place dissolve as you flick through Van Gogh’s Sunflowers and lesser known paintings such as Lippo di Dalmasio’s The Madonna of Humility (where mother and child are bathed in a ring of aureate light).
More academic classifications are there too, with works organised under artists, mediums and movements. But it’s the more ludic aspects of GAC that demand to be considered. Next to the “Explore” tab at the top of the webpage is one called “Play”: features include “Talking Tours”, an AI audio experiment touring cultural landmarks in Street View, and “Mice in the Museum”, involving two rodents (one pink, one brown) scurrying through museum spaces and discussing what they see on the walls.
It’s Arts & Culture meets Walt Disney Studios: but Google’s unfailingly interactive approach poses two, potentially alarming questions. What is the price of constantly gamifying art; and what happens when matters of culture and heritage are filtered entirely through Google?
The systematic gamification of art is a reminder just how much this has come to be viewed simply as a form of entertainment – and how little computer scientists can be trusted with it
Of course, we are some way off GAC becoming the go-to platform for cultural analysis. It seems more likely people will continue to use plain old Google for a while longer. Mr Sood tells me GAC is simply a PR exercise, but if Google’s monopoly on search is anything to go by, then a cultural monopoly cannot be ruled out. Another arm of the company, Google Books, has already proven an invaluable tool for scholars. Eleanor Stephenson, a PhD candidate at the University of Cambridge, says she finds many primary and secondary sources through the website, “all of which have been OCR-scanned”. This allows her to find “crucial passages using keyword or date search” and has “transformed the work of a historian or researcher,” she says.
Many argue the rise of the internet has precipitated a decline in academic and research standards, with a more digitally fluent generation of students disengaging from libraries and source evaluation. If the increasingly capacious offering within GAC helps it become a ubiquitous research tool, it might pose a risk of intellectual complacency among students whose lines of inquiry have been limited by the digital architecture that makes things so seamless. In other words: by making things “too easy”, GAC could limit researchers from digging that little bit deeper and looking for sources outside of it.
Taking Google Books as an example again, archival sources in developing countries are not systematically catalogued, which further distances minority culture from the academic mainstream in an environment where digitised material is, increasingly, the springboard for research. Selectivism is already befalling GAC too, where even certain landmark collections are not displayed on the website. This includes those of the British Museum and the Louvre, safe for a few standout pieces from each.
As for the gamification of culture: is this the future of creativity? Among the GAC features part of its National Gallery collab is a soundtrack function called National Gallery Mixtape, using AI to interpret pigments and shapes in a painting and transposing these to music. Van Gogh’s Sunflowers might generate a track approximating a clarinet concerto; Schiavone’s Jupiter Seducing Callisto might produce something else entirely. Additional “stickers”, including “disco”, “ska” and “chaotic”, allow players to put an extra spin on the composition. It’s a reminder just how much art today is viewed purely as a form of entertainment – and how little computer scientists should be trusted with it.
The audience oohed and ahed during the presentation, but ask me to explain the value of a gimmicky, AI-powered soundtrack to a timeless painting, and I’m afraid to say: I’m stumped.
AI has been a huge boon to artistic and historical research (GAC used it three years ago to restore disappeared masterpieces by Gustav Klimt), but it feels like it’s too often used in puzzling, quixotic ways. As such, it reminds me of YouTube in 2006 when the platform was still figuring itself out and most people had no idea how to use it. Or what purpose it might end up serving.
Art does not need to have a purpose beyond itself, and I hate to speak of creativity in utilitarian terms. Perhaps that’s exactly what GAC is going for here: gimmicks, for gimmicks’ sake. In truth, AI is also still working itself out – something it may only be able to do through trial and error. Outlandish ideas, such as Virtually Parkinson – the world’s first ever AI podcast, modelled on the voice of the late Michael Parkinson and produced by Deep Fusion Films – could just as easily succeed as they could flop. In time, though, these things find their meaning and their groove.
As I go around the Van Gogh exhibition after being shown the digital remasters of the real paintings now standing before me, I cannot help but marvel at the brushstroke and colouring in these spectacular, three dimensional frames. I’m no AI sceptic, but no level of granular detail on a computer screen could ever replace the magic of human touch. This, for now, should be comfort enough.
