Mian Zhang

Concept / Evidence-Gated AI

Evidence-gated AI means proof comes before high-risk action.

This page gives the plain-language entry point for the whole public evidence field.

evidence-gated AI AI evidence gate AI guardrails proof before action AI safety evidence
Evidence Map Papers Registries Counterexamples

Search Intent

Readers want a non-jargon definition before they enter papers, registries, and technical mirrors.

  • What is evidence-gated AI?
  • What evidence should exist before AI acts?
  • How is an evidence gate different from a policy slogan?
  • Where can the claim be challenged?

Definition

The gate is a record, not a slogan.

An evidence-gated AI system requires high-risk action to pass through public or internal proof fields: claim, support, authority, scope, boundary, receipt, and refusal route.

The public site exposes the research-level evidence and boundaries. It does not expose protected private execution details.

Use

The concept is useful only if it can reject action.

If a gate never stops, narrows, or escalates anything, it is decoration.

A real evidence gate can produce no-action, downgraded language, pending status, or a counterexample task.

Evidence Route

Where the claim can be checked.

This page is an entry point. The claim should be evaluated through DOI records, evidence maps, registries, GitHub/HF technical routes, and public counterexamples.

KindAnchorURLRole
Evidence MapPublic claim and evidence maphttps://mianzhang.org/evidence/Start from supported claims and known boundaries.
Paper IndexDOI and paper status maphttps://mianzhang.org/papers/Use paper-specific DOI records for paper claims.
RegistriesMachine-readable public registrieshttps://mianzhang.org/registries/Inspect claim, evidence, action, and counterexample records.
Challenge RouteCounterexample submission pathhttps://mianzhang.org/counterexamples/Attack overbroad claims through public routes.
ArchiveZenodo portfolio indexhttps://zenodo.org/records/20027295Long-term archive index; cite specific DOI records when available.

Boundary

What this page does not prove.

  • This page does not claim that evidence gates solve all AI safety problems.
  • It does not certify production systems.
  • It does not convert public research artifacts into commercial deployment proof.
FAQ

What is the simplest definition?

Evidence-gated AI requires proof, authority, and scope before high-risk action.

FAQ

What should happen when proof is missing?

The system should stop, downgrade, ask, log, or escalate.

FAQ

Where does this connect to existing pages?

Start with No-Proof No-Action, Proof-Carrying Action, Evidence Map, and Counterexamples.