Advanced Intelligent Discovery

About Advanced Intelligent Discovery

A journal built for machine-led discovery

Advanced Intelligent Discovery is a gold open access journal published by Wiley-VCH GmbH in Weinheim, Germany, as part of Wiley's Advanced portfolio of journals. Its online ISSN is 2943-9981, and it publishes in English. The journal's own description on the Wiley Online Library is concise and deliberate: it is presented as the first gold open access title in the Advanced portfolio dedicated to broad-scope research in machine learning, computational science and artificial intelligence, and as a journal that focuses on both experimental and computational research aimed at discovery and design processes. That framing matters, because it separates the journal from general artificial intelligence titles. Advanced Intelligent Discovery is not primarily a venue for theory in isolation, and it is not a trade magazine about the AI industry. Its organising idea is discovery itself: the use of learning systems, computational methods and data-driven design to find, screen, predict and create things that researchers in chemistry, materials science, biology, medicine, energy and engineering actually want to find.

The journal considers applied and fundamental research in machine learning and computational science, encompassing applications such as screening, databases, computational chemistry, materials design, drug discovery, bioinformatics, computer vision, robotics, energy storage and smart grids, among other areas named by the publisher. Read carefully, that list describes a pipeline rather than a single discipline. Screening and databases belong to the input side of discovery, where candidates are generated and organised. Computational chemistry, bioinformatics and materials design belong to the modelling side, where properties are predicted and structures are proposed. Drug discovery, energy storage, smart grids, robotics and computer vision belong to the application side, where a prediction has to survive contact with a real system. A manuscript that connects two or more of those stages — for example a learning model whose predictions are then tested experimentally, or a database resource that makes a screening workflow reusable by other groups — is the kind of work the journal's stated scope is designed to attract.

Editorial leadership is named publicly on the journal's Wiley page. The editor-in-chief is Matteo Cavalleri, and the deputy editors are Richard Murray and Hashini Thirimanne. Authors should always check the journal's current masthead before submitting, because editorial teams for a young journal can change as its article volume grows, but the public naming of an editor-in-chief and two deputy editors is itself a useful trust signal: the journal operates with an identifiable editorial structure inside a large, established scholarly publisher, rather than as an anonymous website collecting manuscripts.

Advanced Intelligent Discovery is a new journal. Wiley announced it among the new artificial intelligence titles added to the Advanced portfolio in its 2025 portfolio updates, alongside Advanced Robotics Research, describing both as open for submissions as part of an investment in chemistry and artificial intelligence within the physical sciences. Its earliest articles visible on the Wiley Online Library date from June 2025, and by 2026 its issue covers were already being numbered within a second volume year. That youth has two practical consequences for authors and readers. First, the journal does not yet have the long citation history, the mature impact indicators or the decades-deep archive that older Advanced titles can show; any profile that presented it as a long-established title would be misleading. Second, its early volumes are unusually informative, because they show in concrete form what the editors mean by discovery: generative models that predict biological function from a single static image, chemical language models specialised for catalyst design, consensus screening pipelines for candidate binders, reviews of monitoring standards for AI-enabled medical devices, and data-efficient deep learning for battery capacity prediction all appear among its early published work.

Scope in detail: what counts as discovery research here

The journal's scope can be understood through the kinds of questions its papers ask. One recurring question is prediction before measurement: can a model infer a property that is expensive, slow or hazardous to measure directly? Early work in the journal includes a study that infers concealed volume loss in a magnesium alloy from optical corrosion images and surface topography, precisely so that a hazardous cleaning step can be avoided in corrosion inhibitor screening. Another asks whether the long-term discharge capacity of a battery electrode can be predicted from limited early-cycle data, which would shorten physical testing cycles. A third applies structure-activity modelling to tattoo and permanent makeup pigments to link physicochemical properties with absorption, distribution, metabolism and elimination profiles. These are different sciences, but the scholarly move is the same: a learned model stands in for a costly experiment, and the paper's value lies in showing how far that substitution can be trusted.

A second recurring question is design rather than prediction: can a system propose a candidate that did not previously exist? The catalyst-specialised chemical language model published in the journal, built on a transformer variational autoencoder and fine-tuned on catalyst databases, is an example, as is the multistrategy AI-assisted consensus screening pipeline developed to find candidate binders for a protein target lacking a settled structure. Design papers in this journal are expected to do more than report that a model generated outputs. The convincing versions constrain generation with templates or physical knowledge, test candidates against independent methods such as docking or co-folding, and discuss druggability, validity or feasibility in the language of the application domain, not only in the language of machine learning metrics.

A third area is the infrastructure of discovery itself: databases, screening workflows, active learning schemes and evaluation practice. A comparative study of active learning schemes for nanophotonics design in the journal illustrates why this belongs in scope. Choosing an optimisation strategy changes what can be designed at a given dimensionality, so a careful comparison of strategies is discovery research even when it does not itself announce a new material. Similarly, work on graph reasoning and knowledge expansion in language models, and reviews examining how AI systems in healthcare should be monitored after deployment, address the reliability layer of the discovery pipeline: how knowledge is represented, how models drift, and how governance keeps pace. Authors whose contribution is a resource, a workflow or an evaluation should frame it in those terms, showing what other researchers can now discover, screen or verify that they could not before.

Equally important is what falls outside the scope. A purely algorithmic paper with no discovery or design application, a purely experimental paper with no computational or learning component, and an opinion piece about artificial intelligence with no research content would each struggle to fit the journal's stated focus. The publisher's wording joins the two halves — experimental and computational, applied and fundamental — and a strong submission normally honours both sides of at least one of those pairings.

Publication model, access and bibliometric footprint

Advanced Intelligent Discovery is a gold open access journal, which in Wiley's usage means that published articles are freely readable on publication under open licences, with the costs of publication normally supported through an article publication charge or through a transformative agreement between Wiley and the author's institution. The exact charge that would apply to a given manuscript, and whether an institutional agreement covers it, are details that are not stated in the material consulted for this profile and can change over time; authors should confirm the current charge and any waiver or agreement eligibility on the journal's Wiley pages before submission rather than relying on any third-party listing, and this profile therefore records the charge as not publicly disclosed in its sources. What can be stated firmly is the access model itself, because the publisher states it: this is the Advanced portfolio's first gold open access title, and its articles are presented as open access on the Wiley Online Library.

Because the journal began publishing in 2025, its bibliometric footprint should be read as an early signal rather than a track record. In a direct ISSN lookup, OpenAlex records the source under ISSN 2943-9981 with 196 works and 380 citations, a source h-index of 9, an i10-index of 7 and a two-year mean citedness of approximately 1.94 at the time of checking. Those figures were retrieved directly from the OpenAlex source record for this ISSN and are attributed to OpenAlex here; they are aggregator figures, they will move quickly for a journal adding articles every month, and they are not a Journal Impact Factor. The journal had no long SCImago or Journal Citation Reports history to report in the sources consulted, which is expected for a title of this age, and readers should not confuse the absence of those indicators with a judgement about quality in either direction. The honest summary is that the journal is publishing real, citable research at a meaningful early volume — 196 works in OpenAlex within roughly its first year and a half — while its long-term standing remains to be established.

For librarians and research managers, the practical points are the stable identifiers and the publisher. The title is Advanced Intelligent Discovery, the online ISSN is 2943-9981, the DOI suffix pattern visible on Wiley pages uses the journal code aidi, as in the 2026 cover article DOI 10.1002/aidi.202600012, and the publisher of record is Wiley-VCH GmbH, Weinheim. The journal should not be confused with its older sibling Advanced Intelligent Systems, a separate Wiley title with online ISSN 2640-4567 that covers intelligent systems, robotics and sensing; the two titles share a portfolio and a naming pattern but they are different journals with different ISSNs, different scopes and different editorial teams. It should also not be confused with similarly named conference proceedings or book series in intelligent systems, none of which carry this ISSN.

Who should submit, and how to judge fit

The natural authors for Advanced Intelligent Discovery are research groups whose work sits at the meeting point of learning systems and a discovery-driven science: computational chemists using generative or predictive models, materials scientists running high-throughput or active-learning workflows, bioinformaticians building screening pipelines, energy researchers predicting storage performance, and engineers applying computer vision or robotics to experimental discovery. Interdisciplinary teams are particularly well matched to the journal, because its early articles frequently pair a methodological contribution with a domain validation, and a paper written by only one side of that pairing often reads as incomplete against the journal's stated aims.

Before submitting, an author should be able to answer three questions clearly. What is being discovered, screened or designed, in concrete domain terms? What does the computational or experimental method add that a conventional workflow could not do, or could not do as quickly, cheaply or safely? And what evidence shows that the output of the pipeline is trustworthy — an experimental test, an independent computational check, a held-out evaluation, or a comparison against an established method? The journal's own early contents suggest that editors and reviewers value those answers, because the published papers consistently foreground validation: synthetic data augmenting a prediction model is reported with similarity scores, a catalyst model is assessed for task-specific validity, and a screening pipeline is carried through to affinity and druggability discussion.

Readers, in turn, should use the journal as a window on how discovery workflows are changing across fields rather than as a substitute for the specialist journal of any one field. A materials researcher will still read the materials literature; what Advanced Intelligent Discovery adds is the transferable pattern — how active learning, generative modelling or consensus screening performed when moved into a new domain. That cross-domain visibility is the stated purpose of a broad-scope title in the Advanced portfolio, and it is the standard against which this young journal should be judged as its archive grows.

Frequently asked questions

Is Advanced Intelligent Discovery open access? Yes. Wiley describes it as a gold open access journal, the first gold open access title in its Advanced portfolio, and its articles are published as open access on the Wiley Online Library. Is it the same journal as Advanced Intelligent Systems? No. Advanced Intelligent Systems is a separate, older Wiley journal with online ISSN 2640-4567; Advanced Intelligent Discovery carries online ISSN 2943-9981 and a discovery-and-design scope. Who leads the journal? The editor-in-chief named by Wiley is Matteo Cavalleri, with Richard Murray and Hashini Thirimanne as deputy editors. When did it start? Wiley announced it in its 2025 portfolio updates and its earliest articles date from June 2025, so it is a new journal whose indicators should be read as early signals. Does it have an impact factor? No Journal Impact Factor for the title could be verified from the sources consulted for this profile, which is normal for a journal launched in 2025; the OpenAlex figures quoted above are aggregator data, attributed as such, and are not an impact factor. What does it publish? Experimental and computational research aimed at discovery and design, across machine learning and computational science applications including screening, databases, computational chemistry, materials design, drug discovery, bioinformatics, computer vision, robotics, energy storage and smart grids.

Reading the early volumes: patterns an author can learn from

The journal's most-cited early articles, as presented by Wiley, repay close reading by anyone preparing a submission, because they reveal the editorial taste behind the broad scope statement. The graph reasoning article published in June 2025 combines a language-model framework with symbolic abstraction and recursive refinement, and demonstrates it across multidisciplinary applications; its appeal lies in showing a method changing what a system can discover, not merely scoring well on a test set. The review of AI-enhanced surface-enhanced Raman scattering, published the same month, works at the opposite pole of the journal: it synthesises a measurement technique transformed by learning methods, and maps its contributions to biofluid analysis, disease identification, therapeutic evaluation and biomedical imaging. A discovery journal needs both poles — new engines of discovery and authoritative maps of fields those engines are changing — and the presence of both in the first month of publication was an early signal that the editors intended the title to be a working venue, not a manifesto.

Later issues extend the pattern into harder territory. The monitoring-standards review of September 2026 addresses AI and machine-learning medical devices entering clinical practice faster than post-market surveillance standards mature, discussing predetermined change-control plans, drift detection, explainability and workflow-aware governance. That is not a discovery result in the laboratory sense, yet it belongs in this journal for a reason its editors evidently accept: a discovery pipeline that cannot be monitored in deployment will not be trusted, and trust is a condition of discovery in regulated domains. Similarly, the nanophotonics active-learning comparison and the battery capacity prediction paper both treat efficiency as a scientific result in its own right — fewer experiments, earlier predictions, shorter testing cycles — which is a defensible position in a journal whose very title promises intelligent discovery rather than discovery in general.

Two practical lessons follow for authors. First, graphical abstracts and structured presentation are part of this journal's culture: Wiley presents its articles with graphical abstracts that state the problem, the method and the gain in one panel, and a manuscript whose contribution cannot be drawn in that form probably has not yet found its discovery claim. Second, interdisciplinary readability is tested at submission, not after acceptance: a chemist, a biologist and a machine-learning researcher should each be able to state what a paper found and why it matters, even if each would verify different parts of it. Papers written only for the smallest specialist circle, with domain jargon on one side and unexplained modelling choices on the other, cut against the broad-scope design that Wiley built the journal to serve.

For research evaluators encountering the journal on a curriculum vitae, the correct reading is symmetrical. A 2025 or 2026 article here is a peer-reviewed, gold open access article in a Wiley Advanced portfolio journal, with a DOI under the 10.1002 prefix and the aidi journal code, full indexing of its text on the Wiley Online Library, and citation accrual already visible in OpenAlex. It is not yet possible, honestly, to place the journal in a quartile or to quote a mature impact indicator, because the title is too young for those instruments to have meaning. Evaluators who need a quality judgement today should read the article, note the publisher and portfolio, and check the source record by ISSN 2943-9981 rather than relying on any unofficial impact score circulating on third-party sites, none of which is relied on in this profile.