Examine
Distinct AI review passes scrutinize claims, sources, mathematics, methods, and the relationship between models and evidence.
THE EDITORIAL MODEL
Critical depth.
Accountable decisions.
An AI-based review process, built around explicit objections, author responses, and human verification.
Challenge & revise
Authors receive explicit objections and respond point by point. Revisions are checked against the original questions.
Verify & adjudicate
Acting editors check consequential findings, assess the response, and resolve disputes. Human editorial judgment determines the decision.
Publish with a record
The intended publication record documents review rounds, substantive changes, and the contributions of AI systems and human editors.
A scientific review record
The intended process uses advanced AI systems in distinct critical passes, rather than asking authors to nominate reviewers. Each pass should identify a specific claim, the evidence or derivation on which it depends, a consequential concern, and a concrete way to resolve it. Repeated review is intended to improve a manuscript; repetition alone does not establish correctness.
What the review examines
Formal review addresses definitions, derivations, assumptions, internal consistency, identifiability, and the mapping from model to observation. Empirical review addresses design, measurement, controls, uncertainty, model comparison, reproducibility, and the scope of the inference. Source checking must distinguish verifiable references from unsupported citations.
Author response & human judgment
Authors must be able to challenge any AI-generated objection, supply a derivation or evidence, and explain a disagreement. Acting editors verify the findings that matter to the decision, assess unresolved issues, and make recommendations within the journal’s editorial governance. AI output does not constitute an editorial decision.
Confidentiality & informed participation
Manuscripts are confidential submissions. Before AI review is activated, the journal will identify the services and systems used, their data handling and retention, contractual confidentiality safeguards, and the information authors receive. Depositing a file does not authorize unrestricted redistribution or use for model training. The intake service itself does not send manuscripts to an AI provider.
Traceability, corrections & appeals
The proposed record includes substantive review issues, author responses, manuscript versions, and disclosure of AI and human contributions. Authors must have an editorial route to challenge a decision or request correction of a factual error. The publication format of the review history, its redactions, and the appeals procedure remain to be finalized.
Protocol status
This page describes the journal’s intended editorial model, consistent with FIELD Atlas Publishing. The protocol is in development. No review time, acceptance rate, accuracy level, or performance advantage is guaranteed. The journal will publish the operative protocol and publication terms before accepting live submissions.