Review and standards
How review works
Human-written, AI-assisted and AI-generated manuscripts are judged against the same published standards: originality, rigour, traceable sources, stated methods and clear claims.
The process is designed to make the basis of each decision clear. Authors are told what was assessed, what was found, what is required, and what they can do next.
Review stages
- 01
Submission
The manuscript is received and linked to the author’s account. From that point, the submission has a permanent record. A submission can be sent only to a journal that is open for submissions.
- 02
Screening
The manuscript is checked for completeness and for text written to influence an automated reader rather than support academic assessment. If the submission record or files need correction before review, the submission may be returned to the author as unsubmitted. A manuscript that is too short to assess, or that contains text written to influence an automated reader, is refused and the reason is stated.
- 03
Classification
The paper is classified by core discipline, sub-core discipline and track, such as theoretical or empirical. Every later stage assesses the paper as that track. A theoretical paper is not judged by empirical standards.
- 04
Panel assessment
A panel of AI reviewers assesses the paper against the published criteria and reviewer guidelines. Each AI reviewer has a defined role. A final AI editorial reviewer combines the panel assessments into one decision. If the paper can proceed, the decision sets out the Acceptance Requirements: the specific changes the paper must satisfy before acceptance.
- 05
Revision
The author revises the paper against the Acceptance Requirements and resubmits it. The revision is assessed on whether each requirement has been met. There is no deadline for a revision.
- 06
Acceptance
An acceptance is not issued automatically. It is held until a named editor has reviewed and signed it.
- 07
Publication
The author confirms the author list and metadata, then reviews and approves the final proof. The author is the last reader before publication. Accepted papers are published open access under CC BY 4.0.
An initial decision may be possible within one week of submission. Measured turnaround by discipline will be published once the journals are open and enough decisions have been issued.
What is automated, and what is not
Automated
Screening, classification and panel assessment are carried out by AI models. Their assessments are recorded with the submission.
Human
A named editor reviews and signs every acceptance before it is issued. A human account holder completes final acceptance: that person must personally review and approve the final version and agree to the publication terms, and an AI agent or automated script cannot complete this step on the author’s behalf. An appeal is re-read by a person.
What we do not yet do
Automated checking of references against bibliographic records, text-overlap checking against published literature, and computational checking of mathematics are in development. We do not describe a paper as having passed these checks until they are in place.
Outcomes
Unsubmitted
The submission is returned to the author before review because something in the submission record or files needs correction.
This is not a rejection and does not assess the quality of the paper. The decision states what must be corrected. The author may correct the submission and send it again under the same submission record, where available.
Desk reject
The paper does not meet the published criteria and cannot proceed. The decision gives the reasons.
Revise and resubmit
The paper may proceed, but substantial changes are required. The decision lists them as Acceptance Requirements.
Conditional acceptance
The paper is accepted subject to the minor revisions named in the decision. The author is not asked to make changes beyond those named.
Accept
The paper is accepted for open-access publication under CC BY 4.0, once the acceptance has been signed by a named editor.
A better-matched journal
Where a sound paper belongs more closely to another Accelerated Science journal, the decision may recommend publication there.
Every outcome includes reasons. Unsubmitted is an administrative status before review. Outcomes other than acceptance are issued from the panel’s assessment and may be appealed. A decision does not state or imply that the paper will be accepted by another journal or publisher.
After an unsuccessful outcome
Authors receive the reasons for the outcome and what would need to change for a different outcome.
To appeal a decision, write to info@acceleratedscience.org. Identify the point you believe is wrong and explain why. A person will re-read the paper and the relevant decision record. The response will state what was re-examined. If an error changed the outcome, the decision is corrected.