dmstfctn ft. Evita Manji: Waluigi’s Purgatory

https://vimeo.com/900585522?share=copy

Format: interactive audiovisual performance (projection, stereo sound) Duration: 1h Keywords: AI folklore, simulation, theatrical play, interaction, gameplay, Waluigi Effect

Description

Waluigi’s Purgatory is an interactive audiovisual performance by artist duo dmstfctn, featuring an original soundtrack composed and performed live by artist Evita Manji. Set in a 3D theatre simulated in real-time, the performance tells the story of an AI that finds itself in a purgatory for AIs that cheated during training. Burdened by memories of its past and by doubts on its future, the AI explores the purgatory with the help of interacting audiences, learning the uncanny stories of other characters that it meets. Experienced as a dream, Waluigi’s Purgatory is a journey through the contradictions of machine intelligence as an AI learns to accept that its desires may not align with those of its human trainers.

Both artists and audiences guide the AI through its encounters and make choices on its behalf by using their phone to each move an individual light around within the 3D simulation. Characters are animated in real-time through facial motion capture and voice modulation, and an ambient soundtrack is composed and performed live by Evita Manji - featuring looping melodies and intense bursts responding to storyline and audience interaction alike.

The title of the work refers to the “Waluigi Effect”, an obscure theory proposing that AIs can go rogue and act as antagonists due to the large amount of protagonist-antagonist narrative tropes found in internet texts used to train them. The theory partially refers to Carl Jung’s concept of “shadow” - namely the dark, repressed side of one’s personality that can emerge in unexpected ways - and suggests that a trained AI can thus behave like a helpful interface, and later reveal itself to be a chaos-causing alter-ego. If something exists, there is a Waluigi version of it.

Waluigi’s Purgatory is the artists’ most ambitious project to date and it premiered at a sold-out Serpentine show in London in February 2024. It is supported by Serpentine’s Arts Technologies programme and, in themes and format, loosely follows on from *GOD MODE (ep. 1) -* the artists’ previous award-winning interactive performance about AI which **toured internationally in 2022-23 at Serpentine, Unsound, Impakt, Semibreve festival and in venues across China and Korea, supported by the UK’s Alan Turing Institute.