The Many Faces of Creative AI in Musical Composition and Performance

Value of the Information & Strength of the Argument

The talk provides valuable insights into the practical and conceptual aspects of using AI in music composition and performance. Laidlow’s argumentation is solid, grounded in his own artistic practice and research. He effectively demonstrates how different AI approaches can serve creative purposes, from machine listening to latent space navigation. He also raises important ethical considerations, such as data sourcing and the potential for AI to automate human work. The inclusion of audio-visual examples strengthens his points, though the reliance on personal experience limits the generalizability of his claims.

Scientific Rigor, Source Quality, Title Accuracy

Laidlow demonstrates scientific rigor by referencing historical figures and projects (e.g., Hiller, Cope, Xenakis) and contemporary works (e.g., Holly Herndon, Zubin Kanga). He also mentions collaborations with institutions like IRCAM and the Intelligent Instruments Lab. The talk is well-structured and the title accurately reflects the content. However, as a seminar, it lacks formal citations and peer review, and some claims are based on anecdotal evidence. The speaker’s expertise and the inclusion of concrete examples mitigate these limitations.

The title accurately reflects the content, which explores various forms of creative AI in music through historical context and the speaker's own projects.

8/10

The speaker is a recognized composer and AI+ Fellow at King's College London, with a strong track record of interdisciplinary work. The talk is based on his own research and practice, providing concrete examples and references. However, it is a seminar presentation, not a peer-reviewed publication, and some claims are subjective.

Markers derived by PSI from the transcript: the creator did not define chapters.

• 02:00Introduction by Elaine Chew and start of talk
• 05:00Historical overview: musical dice games and early computer music
• 10:00Motivations for AI music: fun, rationalizing creativity, new sounds, profit
• 15:00Discussion of contemporary artists using AI (Holly Herndon, Vicky Clarke, etc.)
• 20:00Introduction to Laidlow's own projects and the data-design-intent framework
• 25:00Tūī project: machine listening and bird song
• 30:00Post-Singularity Songs: language models and vocal cloning
• 35:00Techno-utopia: Rave latent space and stacco instrument
• 40:00Q&A session begins
• 45:00Discussion on human agency and future of AI music

• Ferranti Mark 1 recording (1951) — Earliest known recording of music performed by a computer
• Illiac Suite (1957) by Lejaren Hiller and Leonard Isaacson — Early computer-generated composition
• David Cope's EMI program — Emulation of existing composers
• Athanasius Kircher's Musurgia Universalis (1650) — Mathematical system for composing music
• Ada Lovelace's notes on the Analytical Engine (1843) — Discussion on machine creativity
• Holly Herndon's Proto (2019) — Use of neural network Spawn
• Vicky Clarke's Aura Machine — Machine learning with industrial sounds
• Zubin Kanga's Meta Memory (2023) — Blurring past performances
• Conlon Nancarrow's player piano works — Algorithmic composition for player pianos
• Iannis Xenakis' stochastic music — Use of stochastic systems in composition
• Emily Howard's mathematical compositions — Exploration of mathematical shapes in music
• Ted Moore's Quartet (2021) — Use of FluCoMa for data analysis
• George Lewis' Voyager system — Real-time improvisation with AI
• Jennifer Walsh's Ultra Chunk (2018) — Improvised duet with AI digital twin
• Somax2 at IRCAM — Improvising multi-agent system
• Rave (IRCAM) — Neural audio synthesis algorithm used in Techno-utopia
• Intelligent Instruments Lab at University of Iceland — Collaboration on stacco instrument

• Holly Herndon's Proto — Use of AI in music composition
• George Lewis' Voyager — AI improvisation systems
• Somax2 at IRCAM — Multi-agent improvisation