SIQA classification framework
I use SIQA's three-tier framework to classify every artist in the Artist Index by the nature of their AI involvement. The definitions below are the framework verbatim from SIQA; the application notes are mine.
AI-Assisted
A significant combination of human and AI creative contributions, where a human fulfills one or more core creative roles and AI fulfills others. Examples include human production with AI vocals, human vocals with AI production, human songwriting with AI production and vocals, or AI songwriting with human performance.
Human + AI Hybrid
The artist uses AI to clone and reproduce their own voice. The vocal performance is AI-generated, but the voice model is derived exclusively from the submitting artist themselves. All other creative elements — songwriting, production, arrangement — may be human or AI-generated. Self-voice cloning is the defining characteristic of this tier.
Fully AI-Generated
The music — including melody, lyrics, production, and arrangement — was generated entirely by AI. The human role is to prompt, direct, and select the output. No traditional instrument was played, no vocal was performed, and no arrangement was manually composed by the submitting artist.
How I apply it
- Songwriting is decisive. When an artist writes the original lyric source themselves — diary entries, poems, written drafts — I classify the work as AI-Assisted, even when ChatGPT structures the lines and Suno generates the music and vocal. Songwriting is a core creative role.
- Hybrid is specifically self-voice cloning. The defining marker is the artist using a model trained on their own voice (Grimes via Elf.Tech, Holly Herndon via Holly+). Cloning someone else's voice without consent is a different category and falls outside SIQA as currently scoped.
- Fully AI-Generated is reserved for prompt-and-select workflows. No original lyric material from the human, no original musical idea — only direction and curation of model outputs.
- Predominant tier rule. If a project moves between tiers across releases, the infobox carries the predominant tier; per-release exceptions are noted in the Artistic style section.
About SIQA
SIQA is an external framework I adopt as the standard taxonomy for AI involvement in music. The canonical definitions live at thesiqa.com. This page mirrors the three tiers verbatim for use on the Artist Index, and links back to the source for the full framework documentation and any future revisions.