AI

About AI

AI — the MDPI journal with a one-word title

AI (ISSN 2673-2688) is an international, interdisciplinary, open-access scholarly journal devoted entirely to artificial intelligence, published by MDPI in Basel, Switzerland. The Journals World record for this post carries exactly that ISSN and names Multidisciplinary Digital Publishing Institute as publisher, so despite a title that could mean almost anything, the record is pinned to one specific journal: MDPI's AI, whose home is mdpi.com/journal/ai and whose editorial office can be reached at ai@mdpi.com. That disambiguation matters. There is a long-established Elsevier journal called Artificial Intelligence, there are journals called AI & Society, Applied Artificial Intelligence and many others with AI in the name, and there is a large family of newer MDPI titles built around artificial intelligence in a single field. None of those is this journal. If a citation, a DOI or a submission page does not connect back to ISSN 2673-2688 and to MDPI, it does not belong to the journal profiled here.

MDPI describes the journal's purpose in deliberately broad terms: it publishes original research articles, reviews and communications that offer substantial new insight into any field of study that involves artificial intelligence. That breadth is the defining editorial choice. Where a specialist title might restrict itself to, say, medical imaging or natural language processing, AI was set up as a general venue for the whole discipline — theory, technique and application together — at a moment when work on machine learning was expanding faster than the older, narrower journals could absorb it. The journal sits in the early volumes of that expansion: MDPI announced an updated CiteScore for it in June 2025 while it was running in Volume 6, which places its first volume in 2020.

Identity at a glance

• Title: AI
• ISSN: 2673-2688 (electronic)
• Publisher: MDPI (Multidisciplinary Digital Publishing Institute), Basel, Switzerland
• Country on the Journals World record: Switzerland (CH)
• Language and model: English-language, fully open access
• Licence: Articles are distributed under the Creative Commons Attribution (CC BY) licence and authors retain copyright, according to MDPI's copyright statement for the journal
• Article types named by the publisher: Original research articles, reviews and communications
• H-index on the Journals World record: 48 — read the metrics section below before quoting any single figure, because MDPI, Scopus and journal-ranking services report different indicators for this title

Aims, in the publisher's own framing

Two commitments run through MDPI's aims statement for AI. The first is completeness of reporting. The publisher says its aim is to encourage scientists and engineers to publish experimental and theoretical research in as much detail as possible, that the journal places no restriction on the maximum length of papers, and that full experimental details should be provided so that results can be reproduced. In a field where a result often depends on a dataset split, a training schedule, a random seed or a hardware choice that never makes it into a short paper, that no-maximum-length policy is a substantive position, not a formatting detail. The second commitment is openness about supporting material: electronic files and software providing full details of calculation and experimental procedures can be deposited as supplementary materials, and copies of source code or executables, where they exist, may be deposited as well. Readers evaluating a paper in this journal should therefore look beyond the main text — the supplementary record is part of how the journal expects its claims to be checked.

The interdisciplinary framing also deserves to be taken literally. The journal does not present itself as a computer-science journal that occasionally accepts an application paper. It presents itself as a journal for any field of study that involves AI, which is why its published contents range across the areas represented on the Journals World record — advanced neural network applications, AI in healthcare, AI in education, ethics and social impacts of AI, topic modelling, explainable artificial intelligence (XAI), smart agriculture, domain adaptation and few-shot learning. Those record topics are a fair map of the journal in practice: core method work sits alongside domain applications, and alongside papers whose subject is the effect of the technology rather than the technology itself.

Scope, topic by topic

MDPI's published scope list for AI is short and structural. Working through it explains what an editor is likely to regard as in scope.

• AI application, AI technique and AI theory — the three umbrella headings. A submission should be clear about which of the three it principally is, because the evidence expected differs: a theory paper is judged on argument and proof, a technique paper on comparison and ablation, an application paper on whether the AI component genuinely advances the domain problem.
• AI ethics — a named scope area in its own right, covering questions of fairness, accountability, transparency and social impact, rather than an afterthought appended to a technical paper.
• AI planning, including automated planning and scheduling as described in the journal's launch material.
• Artificial neural networks — architectures, training and analysis.
• Natural language processing — language understanding, generation, recognition and related tasks.
• Computer and machine vision — perception, image and video understanding.
• Intelligent machines and agents — systems that perceive, decide and act.
• Machine learning and deep learning — the methodological centre of the modern field, including, in the journal's wider description, knowledge reasoning and discovery and artificial general intelligence as discussion topics.
• Robotics — intelligent robotics in particular, as MDPI's journal-cluster description puts it.
• Genetic algorithms and evolutionary computing — population-based and nature-inspired optimisation.
• Knowledge representation and reasoning — the symbolic tradition of AI, kept inside the same journal as the statistical tradition.
• Expert systems — rule-based and knowledge-based systems.
• Internet of Things (IoT) — intelligent connected devices and the data they generate.

MDPI also places AI inside a publisher-level cluster of artificial-intelligence journals. In that cluster description the journal is characterised as focusing on broad aspects of cognition and reasoning, perception and planning, machine learning, intelligent robotics and applications of AI, and as being led by Editor-in-Chief Prof. Dr. Kenji Suzuki. The same cluster page distinguishes it from narrower sister titles — for example a title devoted specifically to AI in medicine — which is a useful signal for authors: if a manuscript is wholly and specifically about one application domain that now has its own dedicated MDPI journal, an editor may ask why the general journal is the better home, and authors should be ready to answer that question in the cover letter by pointing to the breadth or methodological interest of the work.

Editorial standards and publication ethics

MDPI states that it is a member of the Committee on Publication Ethics (COPE), and its publication-ethics statement for this journal says the publisher enforces a rigorous peer review together with strict ethical policies and standards, treats cases such as plagiarism, data falsification and inappropriate authorship credit very seriously, and uses iThenticate to check submissions against previous publications. Authors coming from fields where AI journals have been slower to formalise these checks should note what follows from that in practice: text overlap with their own earlier papers, with a conference version or with a preprint-derived manuscript can be detected and queried, so the relationship between a submission and any earlier version of the same work should be disclosed clearly rather than left for screening software to surface.

The journal also uses the standard MDPI apparatus of sections and special issues. MDPI's launch editorial for the journal welcomed suggestions for special issues and appropriate guest editors, and special issues have remained a visible part of how the journal organises themed collections. For readers, that means the journal's contents are best browsed both by volume and by collection: a special issue on a narrow topic can contain the most relevant cluster of papers even when those papers span different article types. For authors, it means checking whether an open collection matches the manuscript before submitting to the general stream — and, equally, being cautious about any invitation to submit that does not come from an @mdpi.com address or from the journal's own website, since MDPI explicitly warns, in its fee information, that it does not authorise third parties to act for it.

Indexing, metrics and how to read them

The journal's recent metrics history is documented by MDPI itself in a sequence of announcements that, read together, show a young journal entering the major citation databases. MDPI announced that AI received its first CiteScore of 7.2 in June 2024, following its inclusion in Scopus in 2023, ranking it 100th among 350 titles in the Computer Science (Artificial Intelligence) category for that calculation. A year later, in June 2025, MDPI announced an updated CiteScore of 6.9, ranking the journal 127th out of 450 titles (Q2) in the Artificial Intelligence category, and noted that the journal was then in Volume 6. A fall in a CiteScore alongside a fall in rank is not, on its own, evidence that a journal's papers are being cited less in any simple sense: the category grew from 350 to 450 titles between those two calculations, so the comparison set itself changed substantially, and CiteScore is a four-year average that moves as older, highly cited early papers leave the window.

On its journal homepage, MDPI currently presents a 2025 Impact Factor of 6.5, attributed to Journal Citation Reports (Clarivate, 2026), a JCR quartile of Q1 in Computer Science, Interdisciplinary Applications, and a CiteScore quartile of Q2 in Artificial Intelligence, with coverage listed as including ESCI (Web of Science), Scopus and EBSCO. The Journals World record separately carries an h-index of 48. None of these figures contradicts the others; they measure different things, over different windows, in different databases. The careful way to cite this journal's standing is to name the indicator, its source and its year — for example, that its 2025 Impact Factor as presented by the publisher is 6.5 — rather than to strip a number of its context and present it as a permanent property of the title. Researchers whose institutions evaluate journals by list membership should check the current Web of Science and Scopus source lists directly, because coverage and quartile assignments are revised and a profile page can only report the position its sources showed at the time of writing.

Publication model, fees and speed

All articles in AI are published in full open access. MDPI's fee page for the journal states that an article processing charge of CHF 1,800 (Swiss francs) applies to papers accepted after peer review, covering peer review, copyediting, typesetting, long-term archiving and journal management, with payment also accepted in euros, US dollars, pounds sterling, Japanese yen and Canadian dollars, and with local VAT or sales tax added where applicable. Invoices are emailed shortly after acceptance to the payment contact provided by the authors, and MDPI stresses that only official invoices issued by MDPI are valid. Discounts are described for authors whose institutions participate in MDPI's Institutional Open Access Program. Two practical consequences follow. First, the cost falls after acceptance rather than at submission, so authors should confirm before submitting that a grant, an institutional agreement or another legitimate source will cover CHF 1,800 if the paper is accepted. Second, because the journal is fully open access under CC BY, there is no subscription version of the same article to fall back on — the open-access route is the journal's only route.

On speed, MDPI's journal page reports that a first decision is provided approximately 20.4 days after submission and that acceptance to publication takes about 5.6 days, as median values for papers published in this journal in the first half of 2026. Those are medians for published papers, not promises for any individual submission: papers requiring major revision, additional experiments or a further review round will take longer, and the median for accepted papers says nothing about how quickly a paper that is going to be rejected leaves the system. Authors working to a thesis, funding or tenure deadline should treat the figures as evidence that the journal's production pipeline is fast once a paper is accepted, not as a timetable they can plan around from the day of submission.

Guidance for authors considering AI

The strongest reason to choose this journal is fit at the level of audience. A paper whose contribution is a method, an evaluation or an application that researchers in several AI subfields would plausibly read is better placed in a general AI journal than in a title read by only one subfield. The weakest reason is the title's brevity: because the journal is simply called AI, a publication in it can be misread in a reference list, and authors should make sure their own CVs and repositories give the full citation with the ISSN or DOI so that the paper is not confused with work in Artificial Intelligence or another similarly named venue.

Reproducibility deserves special effort here, because the journal itself makes it an editorial theme. Authors should report datasets and their versions, splits, preprocessing, model configuration, hyperparameters, evaluation metrics and their exact definitions, the number of runs behind any averaged result, and the hardware or software environment where it materially affects the result. Where code or data cannot be released — for example because a dataset is clinical or commercially restricted — the paper should say so plainly and explain the restriction, rather than omitting the topic. Supplementary files and code deposits are welcomed by the publisher, so there is rarely a space argument for leaving methodological detail out; the journal has no maximum length.

Ethics statements matter in proportion to the topic. Work involving human data, biometric data, medical records, children, or automated decisions about people needs the approvals and consent arrangements its own institution and jurisdiction require, described in the paper. Work whose main contribution is a capability with foreseeable misuse should discuss limitations and risks frankly. The journal names AI ethics as a scope area, so editorial readers are likely to notice when a paper that clearly raises such questions says nothing about them.

Guidance for readers and librarians

For readers, the practical skill with a broad, fast-growing journal is filtering. Search within the journal by section and collection as well as by keyword, sort by the article type you need — reviews for orientation, research articles for evidence — and read the supplementary material before relying on a headline result. For librarians and research managers verifying the title, the stable identifiers are the ISSN 2673-2688, the publisher MDPI, the DOI prefix pattern used across MDPI titles and the journal's presence in the Scopus source list and in Web of Science's Emerging Sources Citation Index as presented by the publisher; the one-word title on its own is not a sufficient identifier in any catalogue, and cataloguing it without the ISSN invites exactly the confusion this profile opened with.

Frequently asked questions

Is AI the same journal as Artificial Intelligence? No. Artificial Intelligence is a separate, much older Elsevier journal. AI is MDPI's open-access journal with ISSN 2673-2688.

Is AI open access, and what does it cost? Yes, fully. MDPI lists an article processing charge of CHF 1,800 for papers accepted after peer review, with institutional-programme discounts described on its fee page.

Is AI indexed in Scopus? MDPI reports that the journal was included in Scopus in 2023 and received CiteScores in 2024 and 2025, and its journal page lists Scopus and ESCI among its coverage. Always confirm current coverage in the database itself before relying on it for an evaluation requirement.

Does the journal limit paper length? MDPI's aims statement says there is no restriction on the maximum length of papers, and it asks for full experimental details so results can be reproduced.

Who leads the journal? MDPI's artificial-intelligence journal cluster description names Prof. Dr. Kenji Suzuki as Editor-in-Chief of AI. Editorial leadership changes over time, so check the journal's editorial board page for the current holder before addressing correspondence.

What is not publicly disclosed here? Acceptance rates and a complete year-by-year editorial history are not published in the sources used for this profile, so they are not stated. Where a figure could not be verified from an official or indexing source, this profile leaves it out rather than estimating it.