Using AI to Produce “Impossible” Tulips

Anna Ridler, "Myriad (Tulips)" (ⓒ 2019 Anna Ridler)
Anna Ridler, “Myriad (Tulips)” (ⓒ 2019 Anna Ridler)

BERLIN — Reaching a fever-pitch in the 1630s, “tulipmania” — a Dutch Golden Age obsession with the rare and exotic flowers responsible, supposedly, for driving overzealous buyers to financial ruin — has long been considered the first economic bubble. The tulip craze served as a convenient analogy for stories of our desire to monetize the natural world and our tendency towards speculative absurdity. While the extent of this botanical craze has been vastly exaggerated in books, blockbuster movies, and principles in economics, the idea that flowers might control markets continues to captivate social scientists as well as artists. In her latest work, London-based artist Anna Ridler brings this historic phenomenon into the future, using AI to produce thousands of invented “impossible” tulips, slowly developing the features that early modern collectors considered valuable — their unpredictable stripes and stipples — along with the price of bitcoin.

A screenshot of “Mosaic Virus,” visit Vimeo to watch the video

Ridler’s video installation, “Mosaic Virus,” is named for the plant virus that creates the strange variations in color that catapulted the price of some tulips far beyond others for 17th century collectors. Commissioned by Netherlands-based arts organization IMPAKT, the installation is currently on view as part of ERROR — The Art of Imperfection, an exhibit at Ars Electronica Export in Berlin. Using a generative adversarial network (GAN) algorithm trained on thousands of the artist’s own floral photographs (themselves forming a separate installation, “Myriad (Tulips)”), what emerges is a hypnotic grid of artificial, mostly pink and white false blooms mutating as bitcoin markets rise and fall, occasionally reaching stasis. While some flowers appear deceptively realistic, others split and overproduce, morphing into unrecognizable if decidedly still floral forms. The result is breathtaking — a stunningly prescient critique of value, human control, and the tenuous line between “nature” and “artifice.”

AI-generated art like Ridler’s has stoked significant controversy in the past year, culminating in the questionable sale of another GAN-based portrait for more than $430,000 at Christie’s. While some artists seem to delight in the question of whether algorithms can produce work of equal (or higher) value to humans, Ridler defines her work in AI as “craft,” comparing her practice to the production of tulip-filled Dutch still-lifes. Trained at the Royal College of Art and Oxford University, she notes that the division between work and authorship is as contentious in AI as it has been, historically, in both botany and workshop painting. While algorithms are typically associated with individual or institutional authors, Ridler acknowledges that it’s easy to overlook “the labor that goes into making a dataset,” ignoring the network of “human decisions” involved in machine learning — that data “always starts out as objects in the real world before being abstracted and passed into an algorithm.”

Anna Ridler, "Myriad (Tulips)" (ⓒ 2019 Anna Ridler)
Anna Ridler, “Myriad (Tulips)” (ⓒ 2019 Anna Ridler)

Ridler’s commitment to making her own hand and careful eye visible in her work differentiates “Mosaic Virus” from most other algorithmic art. Displayed next to its photographic training set, the dizzying, grid-like “Myriad (Tulips),” “Mosaic’s” painstaking human labor counteracts the supposed “objectivity” of artificial intelligence. Spending spring through early summer hunting through Dutch flower markets, the artist hand-picked more than 10,000 (in classical terms, a myriad) “ideal” flowers. She then isolated, photographed, labeled, and arranged her subjects according to color, striation, species, and state of decay. The artist’s “fieldwork” traced the cycle of the tulip’s bloom season, ending, naturally, when peonies overtook pride of place in flower markets.

Native to the valleys of the Tien Shan Mountains and cultivated across the Ottoman Empire by the 11th century, tulips have always been objects of desire thanks to their fantastic variations in color. After being transplanted to the Low Countries in the 1600s, the flowers took on a life of their own. Sold for ever-increasing amounts in Dutch marketplaces and auctions, alongside other “exotic” collectibles like shells and fine art, tulips demanded vast speculative investment from their buyers. Traded as yet-unblossomed bulbs, collectors paid for what the flowers might become during their short bloom-cycle, devoting anywhere from a day’s wages to years’ worth of fortunes to future floral products. These fragile negotiations between risk, trust, and floral ephemerality drove the eventual market crash, although, as historian Anne Goldgar has deftly argued in Tulipmania: Money, Honor, and Knowledge in the Dutch Golden Age, claims that the tulip trade triggered something akin to a social and economic collapse has little grounding.

Ridler is not the first to draw parallels between this object-based speculation and the sudden rise and risk of cryptocurrency, but the connections between tulipmania and bitcoin have drawn significant skepticism — particularly from Goldgar herself. When displayed alongside “Myriad (Tulips),” Ridler’s work only serves to strengthen the constructedness of “crazes.” Injecting the “hands-on-ness” back into both machine learning and cryptocurrency markets, the artist humanizes processes that can seem coldly removed from the object-based trading that took place on 17th century docks, in flower markets, and in illustrated catalogues.

In her careful process of selection, arrangement, and repetition, Ridler’s work mirrors both horticultural craftsmanship and the process of still-life painting so characteristic of the Dutch Golden Age. Depicting flowers in permanent bloom and decay with exaggerated colors and unnatural groupings, these still-lifes represented what Ridler calls, in her artist’s statement, “botanical impossibilities” — groupings of imagined and constructed naturalia that, in fact, defied natural limits. Bouquets of perfectly formed and radically striped tulips, frozen in time, never existed at all.

After all, the problem with tulipmania lay somewhere between timing and in the limits of human rule over nature. Collectors invested in bulbs months before they bloomed and exerted little control over tulips’ coloration, the foremost marking of their value. Indeed, as Ridler noted in categorizing her own training set, neatly differentiating between various hues — white and pale pink, dark pink and red proved impossible for both the artist and for her algorithm. “Mosaic Virus” and “Myriad (Tulips)” make the case that the artist and algorithm can never be separated, that flowers are inextricable from systems of social and economic currency.

Mosaic Virus and Myriad (Tulips) continue through March 3 as part of Error—The Art of Imperfection at Ars Electronica Export (Berlin). Mosaic Virus will be on view from May 16–August 26 at the Barbican’s seasonal show, AI: More Than Human (London). More of Ridler’s work and additional information can be found on her website.

The post Using AI to Produce “Impossible” Tulips appeared first on Hyperallergic.

Original source: https://hyperallergic.com/487261/anna-ridler-tulipmania/

Leave a Reply

Your email address will not be published. Required fields are marked *