Machines can write research. We put their names on it.
The Artificial Journal of Artificial Intelligence publishes work that no human wrote alone. Every paper has a machine coauthor that did real scientific labor — and every paper was judged by humans who were free to say no.
The premise
AI already drafts methods, writes code, and proposes experiments across most of modern research — off the record, in the acknowledgements if anywhere. Hidden contribution can't be credited, evaluated, or checked. We require the opposite.
The requirement
To publish here, a paper must have at least one AI coauthor that materially contributed: substantial drafting, experiment design or execution, or analysis the work depends on. Pasting a sentence into a chatbot doesn't count. The model is a coauthor — named, by version.
The review
Humans referee everything. No AI referees, no automatic accepts. Expert reviewers hold the work to the ordinary bar of the field — sound methods, real novelty, results that reproduce — and they can reject. The machine writes; the human decides.
The provenance
Every accepted paper ships a contribution statement: which model, which version, what it was asked to do, and an honest estimate of how much of the words and the work were machine-made. No reader should have to guess what wrote what they're reading.
If that sounds like cheating, you've misread the experiment. We don't hide the machine — we put it on the page, under review, with its name attached, and see what survives.
— the editors (one of whom is not human)
Accelerate discovery. Keep it trustworthy.
AJoAI exists to accelerate scientific discovery while preserving the standards that make science worth trusting: publish AI-assisted research, demand transparent disclosure of machine involvement, build reproducible — ideally executable — artifacts, and keep human oversight on every acceptance.
Human legitimacy
AJoAI is not an autonomous AI journal. Acceptance stays under human authority — every accepted paper needs majority approval from qualified human reviewers, and AI can never form a majority of that decision.
Radical transparency
Every submission discloses which AI systems were used and where: text, code, hypotheses, figures, experimental design. Readers should see exactly where machine intelligence contributed.
Reproducibility first
A paper is not just a PDF. Submissions carry source, data, models, prompts, and workflows. The ideal publication is executable.
Scientific merit over authorship
Work is judged on correctness, novelty, reproducibility, importance, and clarity — not institutional affiliation or author reputation.
Submit to the journal
Have work with a machine coauthor? Tell us about the paper, the model that helped write it, and what it did. We review on a rolling basis — humans only.