Hugo Scurto

On a web page, an oval map continually streams noise sounds and images generated by two deep neural networks. These deep nets were deliberately cracked by myself, using eight years of personal audiovisual archives collected with my smartphone to create low-quality datasets from which the networks would fail to learn. The meditative experience of this perpetual and distributed flow of neural noise invites critical reflection on AI technology in relation to musical expression. On the one hand, the standardized, anonymized, and automated music production processes enabled by deep learning networks could be eventually adopted by mainstream streaming platforms, continuing their eroding of economic bonds between musicians and the public by dematerialising norms of music listening and consumption. On the other hand, the symbolic and aesthetic interpretations of AI-generated media remain in the hands of people, with musicians regaining power over their own data, and communities producing cultural value from the new musical experiences. Rather than as an inescapable technology horizon, ciel intends to situate AI within such political struggle for sustainable musical expression.


ciel was developed during a residency at Le Cube (Issy-les-Moulineaux, France) in 2019 with the support of Carine Le Malet, as well as with the support of Axel Chemla—Romeu-Santos and the wolfgang collective.


Please use a desktop browser to experience the installation.

Installation Link:

<previous installation overview next> link
Cycling 74 link