AI Atlas Mission and Vision

One-line identity

AI Atlas is an open, community-reviewed knowledge map of Artificial Intelligence, built on a structured taxonomy.

Mission

AI Atlas helps people understand AI by mapping stable concepts, their meanings, and their relationships.

Vision

AI Atlas aims to become a trusted conceptual navigation layer for AI, helping engineers, researchers, educators, builders, and decision-makers see how AI concepts connect across fields, systems, architectures, tools, and real-world applications.

North star

AI Atlas focuses on durable concepts and meaningful relationships rather than product lists, temporary market labels, or isolated definitions.

Understanding AI is the mission. Education is an important use case.

The taxonomy is the structured foundation. The knowledge map is the long-term product vision.

Taxonomy and knowledge map

AI Atlas uses a structured taxonomy as its source-of-truth foundation. The taxonomy organizes concepts by broader–narrower hierarchy.

The long-term knowledge map extends beyond hierarchy by also representing meaningful relationships between concepts, such as conceptual overlap, technical lineage, dependency, implementation, evaluation, or application.

Hierarchy and relationships are different concerns. AI Atlas should not force cross-cutting concepts into a single parent branch just to express non-hierarchical relationships.

Concept modeling principles

• Model concepts before artifacts.
• Prefer stable, academically recognizable, technically useful terminology.
• Treat papers, repositories, products, benchmarks, companies, datasets, and model releases as evidence or examples unless they are explicitly in scope as canonical concepts.
• Avoid temporary market labels, product branding, or concrete model versions unless clearly scoped and reviewed.
• Use external taxonomies, ontologies, and knowledge graphs as references, not as automatic sources of canonical AI Atlas structure.

Current public scope

AI Atlas v0.1 is a public preview of the Level 1–Level 2 taxonomy.

Level 3 material is not canonical until it has passed the project review and promotion process.