'Artificial Intelligence?' No, Collective Intelligence.

May 24, 2024

Guests: Holly Herndon

Holly Herndon is an artist and musician who has spent years making AI-centered music, including her 2019 album PROTO (created with the AI voice model 'Spawn' trained on a human chorus) and 'Holly+,' a voice model trained on her own voice that she opened for public use; she co-founded Spawning, a project building opt-in/opt-out infrastructure for artists whose work is used as AI training data.

Listen to the episode

About this episode

Klein argues most AI art is 'crappy' because generative systems are inherently generic, then presents Holly Herndon as a counterexample: she uses AI to become 'weirder, stranger, more uncanny, more personal.' Herndon — raised singing in church choirs in East Tennessee before moving through Berlin techno, which she calls a 'coordination technology' — explains PROTO (2019), built with 'Spawn,' an AI 'baby' trained on human ensemble voices, and her later Holly+ voice model, a governed public instrument built from her own voice. Her core claim: AI is not artificial intelligence but collective intelligence — 'aggregate human intelligence, trained on all of us' — more like choral singing than a machine oracle. They explore the meaning question (why AI outputs feel hollow), her Spawning project building universal training-data opt-out tools, the flattening sameness of infinite media, and her insistence that 'the model is the art.'

Read the transcript

Books recommended

0

No book recommendations recorded for this episode.