Open Data for AGI: Why It Matters

AGI will be shaped less by who has the best architecture and more by who has the best data — and whether that data is open. A model trained only on proprietary, uninspectable corpora inherits every bias, gap, and licensing risk of its inputs, with no public record to audit.

Three reasons openness is load-bearing

• Robustness: public datasets get probed, corrected, and versioned in the open. Errors surface faster than in any closed QA process.
• Fairness: you cannot measure representation gaps in data you cannot see. Open corpora make audits possible.
• Trust: high-stakes deployments need provenance. "Trained on public data, version X" is a claim anyone can verify.

What this means in practice

Prefer reference data with clear provenance and permissive licensing — this catalog is dedicated to the public domain (CC0) precisely so agents and models can consume it without legal friction. When you publish data, attribute sources cleanly so the chain of provenance stays traceable.

Go deeper

This note summarizes the argument of Open Data for AGI, a practical brief covering the major open datasets, the legal and economic forces shaping data policy, and recommendations for engineers, researchers, and policymakers. The catalog gives you the data; the book gives you the strategy.