Unlike conventional taxonomies or tool lists, this Compass offers a holistic view of the AI-art space, provides a compact and stable skeleton for understanding a complex and rapidly evolving landscape, and serves as a practical design tool for structuring and reconfiguring AI-art practices. It provides: (1) a single, unified space in which any AI-art project can be situated and succinctly characterized; (2) an actionable design space that enables deliberate moves along integration depth, artwork facet, and value orientation rather than simply adding more models; and (3) a structural language for evaluating and justifying AI-art practices in terms of what they change and which kinds of value they are meant to pursue. 1. A common placement for any AI-art project Any project can be written as: Position = (Depth of AI Integration, Artwork facet, Value direction) Mechanism = combination of AI Core Functions used on that facet. Depth of AI Integration reuses the Four-Tier Framework for Human-AI Collaboration [DOI: 10.5281/zenodo.17701769]. It forces you to say how far you choose to embed AI: from external assistant, through workflow engine and domain adaptation, up to system-level redesign. Artwork is split into Meaning, Form, Context, so you must say whether AI is touching expression (concepts, narratives, emotions), sensory/material realization, or situations/systems/audiences. The Value Compass (Aesthetic, Creative, Socio-cultural) makes you state which notion of “good” the project primarily serves. The AI Core Functions come from the AI Core Function Ontology [DOI:10.5281/zenodo.17664037] and provide a model-agnostic vocabulary for how a project works: they distinguish whether you are mainly representing, reasoning, optimizing, simulating, generating, or orchestrating on a given artwork facet. Decomposing a project as a combination of these functions gives you a concise, comparable, mechanism-level profile, independent of specific architectures, models, or tools. Because every project can be placed in this 3D space and expressed as a Core-Function combination, you get a compact, comparable “profile” instead of a pile of bespoke descriptions. 2. A design space you can move in The central AI–Art Regime Map is the 4×3 grid of Integration × Artwork. Each cell is a usage pattern (a regime), independent of model names. Colors indicate where practice is dense vs sparse. Because the axes are explicit, you can ask very concrete design questions: Integration move: “What if this idea went from ‘tool’ to ‘embedded workflow’ or ‘domain-adapted collaborator’?” Artwork move: “What if we apply the same Core Functions to Meaning instead of Form, or to Context instead of Meaning?” Value move: “What if we re-shape this pattern to target Creative or Socio-cultural value, not only Aesthetic?” 3. A structural language for evaluation and justification Because each project has both a Position (Integration × Artwork × Value) and a Mechanism (Core Function combination on a facet), the Compass gives you a way to argue about AI-art work, not just describe or locate it. You can make precise statements such as: “This project only operates on Form at shallow integration tiers and relies purely on generation; its impact is mainly aesthetic and surface-level.” “This project operates on Meaning or Context at higher tiers and combines reasoning, simulation, and orchestration; it has the potential to restructure concepts, narratives, or institutional relations.” In other words, the Compass gives you a structural language to say what is being changed, how deeply, by which generic abilities, and toward which notion of ‘good’, so you can justify whether a given way of using AI in art is trivial, risky, or genuinely significant. Responsible use note: This Compass is a decision-support framework, not a normative standard or automated decision rule. You should adapt it to your context, values, constraints, and legal/ethical requirements, and remain accountable for downstream choices and impacts.