Tyranny of (AI)Thought. Artificial intelligence in music composition: a case study on Krallice “Diotima”

In this article, we intend to discuss the growing influence of artificial intelligence (AI) in the fields of art. The phenomenon, which has long been familiar in the visual arts sector, is now also gaining relevance in the mu-sic sector. The impact of AI on music, particularly on the more avant-garde and experimental sub-genres, is generating debates ranging from copyright, authenticity and, above all, creativity. The specific case dis-cussed in this one is the album Diotima by New York band Krallice, released in 2011, and its ‘counterpart’, Coditany of Timeness, generated by Dadabots, a collective experimenting with AI, created using a neural network trained on Diotima. Released at NeurIPS 2017, Coditany of Timeness exemplifies how AI can reinterpret musical styles, enabling a reflection on ‘machinic’ creativity. Both albums were subjected to analysis with regard to song structure, melody and harmonic language. The creative dimension behind the project was deepened through a textual analysis of articles and online interviews, a total of 37 documents. This set of results was then discussed with members of the Krallice through a semi-structured interview. The intention is to explore how AI interacts with the human creative process, asking questions about how the machine interprets these processes, and how these results are perceived by the artists themselves.