--- name: nature-machine-intelligence description: Use when targeting Nature Machine Intelligence (Nat Mach Intell) or deciding whether an AI, ML, or robotics manuscript fits this venue. Encodes the journal's fit, framing, method-and-evidence bar, house style, official-submission re-check, and desk-reject heuristics. --- # Nature Machine Intelligence (nature-machine-intelligence) ## Journal positioning Nature Machine Intelligence is a Springer Nature journal publishing research across machine learning, artificial intelligence, robotics, and their intersections with other sciences and society. It occupies the Nature-family tier, so the bar is not a strong ML method alone but a result with conceptual significance or real-world relevance that will interest researchers across AI and the sciences more broadly. The journal explicitly attends to societal impact, ethics, fairness, and the responsible development of AI — these are not boxes to tick but genuine editorial concerns. Papers that demonstrate a compelling application of AI to a scientific or societal problem alongside methodological rigor are a strong fit. This skill is a **fit / venue-selection / re-framing** tool. It does not replace the journal's current official submission guidelines. Before submitting, re-check the live author instructions on the Springer Nature site and the submission system. ## When to trigger - The author names Nature Machine Intelligence or Nat Mach Intell as the target venue. - An ML or AI paper has strong conceptual novelty or a major real-world application and the author is choosing between this venue and specialized venues. - A paper applying AI to scientific discovery (biology, chemistry, physics, climate) needs framing for an interdisciplinary Nature-family audience. - The author wants to assess whether societal/ethics dimensions are adequately addressed and what desk-reject risks exist. ## Scope & topic fit - Novel ML/AI methods with demonstrated conceptual advance or substantial real-world significance — benchmarking on existing tasks alone is insufficient. - AI applied to scientific discovery: protein structure, drug design, materials discovery, climate modeling, astrophysics, genomics — where AI enables a result inaccessible by prior methods. - Robotics and autonomous systems with broad significance and demonstrated capability in challenging, realistic settings. - Human-AI interaction, explainability, and fairness research grounded in rigorous methodology. - Computational neuroscience and cognitive modeling at the intersection of ML and brain science. - Analyses of societal, ethical, or policy dimensions of AI, when empirically grounded and methodologically rigorous. ## Method & evidence bar - Novelty must be conceptual or applicative at the Nature-family significance level: a marginal gain on a leaderboard does not clear the bar; a new learning paradigm, a structurally different architecture with principled justification, or a result enabling new scientific insight does. - Empirical claims must rest on rigorous evaluation: appropriate baselines, multiple datasets or settings, statistical testing of performance differences, ablation studies establishing the source of improvement. - Code and data availability are strongly expected; models and datasets should be released or clearly committed to release; re-check current Nature Portfolio norms. - Ethical considerations must be addressed substantively: training data biases, dual-use risks, fairness across demographic groups, potential for misuse — where applicable. - For AI-in-science papers, the scientific result must be validated and the AI component should not simply serve as a black-box accelerator without interpretive insight. ## Structure & house style - The opening must frame why this advance matters beyond the ML community — what scientific, social, or technological barrier it lifts. - Nature-family format: concise main text (Article or Letter); Methods section at end of or after main text; Extended Data for additional experiments and supplementary information for ancillary material. - A Nature reporting summary (for life sciences / methods-dependent work) may be required; re-check current requirements. - Figures must be self-contained and interpretable by a non-specialist: clear captions, defined notation, illustrative architecture diagrams. - Ethics statement and data/code availability statement are required; re-check current Nature Portfolio standards for AI disclosure. - Competition with recent advances: the introduction must position the paper against the state of the art explicitly, not claim novelty through omission. ## Official-submission checklist - Before giving submission-ready advice, read `../../resources/source-basis.md` and `../../resources/official-source-map.md`; start from the official source anchors for this journal family, then cite the current journal-specific page you checked. - Search the live site for "Nature Machine Intelligence author instructions" and follow the current Springer Nature version. - Re-check article type (Article, Letter, Review, Perspective, Comment), length limits, and abstract format. - Confirm code and model availability commitments; check for Nature Portfolio data-sharing policies. - Prepare reporting summary if required for the study type. - Complete ethics statement, competing-interest declaration, author-contribution statement, and AI-use disclosure (for manuscript preparation). - Confirm preprint policy and open-access / licensing options. - If the live official instructions conflict with this skill, the official instructions win. ## Pre-submission self-check - [ ] One sentence stating why this AI/ML result matters to researchers beyond the immediate ML subfield. - [ ] The contribution is stated as a conceptual advance, a new capability, or a scientific discovery enabled by AI — not as a benchmark improvement alone. - [ ] Evaluation covers multiple baselines, settings, and ablations; statistical significance is addressed. - [ ] Ethical considerations, dual-use risks, and fairness dimensions are substantively discussed. - [ ] Code, models, and data are committed to open release or a clear justification is given. - [ ] Extended Data, supplementary information, and reporting summary are prepared per current instructions. ## Common desk-reject triggers - A strong method paper with state-of-the-art results on standard benchmarks but no conceptual novelty or broader significance beyond the ML community. - Missing or perfunctory ethics discussion for a paper with clear societal or dual-use implications. - No code or model release and no compelling justification for the omission. - The AI-in-science paper delivers a better prediction but not a new scientific understanding. - Incremental extension of an existing architecture without principled motivation or demonstrably new capability. ## Re-routing decision Excellent ML methods papers that are rigorous and complete but narrower in scope → `journal-of-machine-learning-research` or `ieee-transactions-on-pattern-analysis-and-machine-intelligence`. Robotics-specific work with demonstrated capability → `science-robotics`. Fundamental ML theory → `journal-of-machine-learning-research`. Broad-audience conceptual AI breakthrough → consider Nature or Science. ## Output format ```text [Fit] High / Medium / Low (one-line reason) [Target] Nature Machine Intelligence [Topic tags] <2–3 closest topics> [Method/evidence] [Top risk] [Official items to re-check]
[Re-route suggestion] ```