These historical artefacts are totally faked

    30
    0

    These historical artefacts are totally faked

    Nora Al-Badri was bored by deepfake porn. She thought the technology, best known for putting people’s faces into videos they weren’t actually in could be put to work doing something better. As an artist who regularly works with digital technology, Al-Badri had an idea for a more interesting project employing an AI technique known as a generative adversarial network (GAN), commonly used for deepfakes.

    That work, Babylonian Vision, used GANs to expand what we know about ancient history and question museums’ ownership of objects in the 21st century. Al-Badri, who is German with Iraqi roots, took 10,000 photos of Mesopotamian, Neo-Sumerian, and Assyrian artefacts from five museum’s websites and used the images to train a GAN that then created 200 made-up pictures. The work, which was showcased at EPFL Pavilions in Switzerland last year, is dream-like, with misshapen vessels, vaguely human-shaped statues, and incomplete pieces of jewellery, but still clearly echos the art in the dataset it was created on.

    Although Al-Badri said she asked permission from each museum beforehand, only two – the Met and the Cleveland Museum – had APIs that made it easy and clear how she could use the photos. The other three museums either charged or had onerous rules to get them at the scale Al-Badri needed them, with one telling her that she would have to ask for permission for each photo of the thousands she needed. For those institutions, which Al-Badri has declined to name, she set up a web scraping tool to download the pictures she needed. “They don’t understand their digital collections,” she says. “They also probably don’t understand AI.”

    Al-Badri calls her use of technology on cultural artefacts “technoheritage,” an approach more interested in exploring the future than looking at the past. It’s not preserving some piece of history; it’s offering a new starting point for artists to extend our understanding of an object. This type of work could become increasingly widespread, especially after the pandemic, when many museums beefed-up digital collections to engage with the public.

    With putting more work online, museums will have to grapple with artists using their collections in new ways, taking control of the narrative out of the hands of traditional curators. New technology like AI will not only create additional questions already ongoing around the discussion on museum ownership, but could create novel issues in the art world that need to be sorted out. The work that comes out could be worth it though. “It’s a great application for museums in some ways,” she says. “If you have knowledge and you share it, you learn too.”

    Babylon Visions is a creative way of filling in the historical record, but it’s also a strategy to reclaim a culture. Many museum collections rely on artefacts from colonised countries, some of which are currently working on getting pieces returned. One of Al-Badri’s best-known projects was making a 3D scan – exactly how she did it is contested – of the famous Nefertiti bust housed in the Neues Museum in Berlin.

    Published at Sun, 24 Oct 2021 05:00:00 +0000

    https://www.wired.co.uk/article/fake-artefacts-ai

    Previous articleSupercar Makers Skid Into EV Era
    Next articleMyanmar says it’s committed to ASEAN peace plan, despite general’s snub