Today, the Mapping the Arts and Humanities Project (MAHP) at the School of Advanced Study (SAS), University of London, launches Mapping the Humanities and AI resource: a live, searchable map of the 316 centres, institutes, labs, networks, learned societies, associations, and grassroot collectives working at the intersection of AI and the humanities.

AI is too often treated as the property of computer science and engineering. As the Edinburgh Futures Institute (EFI) notes, “for most of its history, AI’s lingua franca was numbers,” but “technical expertise alone cannot address AI’s most critical challenges” (Hemment and Kommers, 2025). SOAS echoes this, adding that “the logic and reasoning behind Artificial Intelligence… is the humanities,” given that humanistic disciplines have spent centuries examining the boundaries of non-human knowledge, ethics, and personhood (Imafidon, 2026). 

The map makes visible the varying degrees to which the humanities interrogate and shape AI. It spotlights the humanities-led infrastructures that are driving how AI is developed, governed, legislated, and used. Crucially, it also maps STEM-focused institutions that are actively bringing humanities-based questions into their labs to tackle complex issues like algorithmic bias and digital rights.

We see this with centres from the Data Justice Lab at Cardiff University to the Centre for Dance Research at Coventry University, ranging from the relationship between datafication and social justice to robots alongside dancers as a way of archiving performance. We see it at Oxford’s Institute for Ethics in AI, where philosophers work with scientists, industry, government and civil society. Its laboratory is devoted to “philosophy to code”: an open-source pipeline that puts AI into the service of human flourishing. And we see it at home: the London AI and Humanity Project at the School of Advanced Study’s Institute of Philosophy pairs philosophers with industry researchers investigating how people interact with AI, runs a summer school with Hong Kong University that has travelled from Paris to Venice, keeps a glossary of the field open to all, and has briefed civil servants at DSIT and the Cabinet Office. The questions AI poses are transnational and cross-sector, and these infrastructures show how the answers cross borders and sectors too.

The map is an exciting milestone because it showcases the sheer scale of this interdisciplinary ecosystem. Unlike a fragmented internet search (which often returns disjointed results and lacks the capacity to reliably demonstrate broader themes or aggregate trends), the (H)AI Map reveals the hidden connections and overlaps across 316 distinct hubs, offering an unprecedented, holistic view of how humanists are redefining artificial intelligence.

The resource enables you to search, filter, and explore this wealth of infrastructural knowledge in one place: start here.

➡️ Read our previous blog post, “In Development: Mapping the Humanities and AI,” for the thinking behind the map and the wider field it sits within.

➡️ Read “Mapping Openness and AI,” the first in a series drawn from the BRAID “AI and Openness” workshop, for the questions of openness, consent, and governance that run underneath this work.

Built on two maps’ worth of lessons

The (H)AI Map is the third subject map MAHP has produced, after the Law and the Humanities Map (LHub) and the Knowledge Diplomacy Mapping Initiative (KDMI). The live, self-updating dataset and the relational visualisations were refined across both earlier maps; KDMI’s first experiment in surfacing funding has grown here into a full Funded Research panel; and LHub’s much-used entanglement idea returns doubled and rebuilt, as an AI score and a humanities score. This time, the AI score looks for the specific methods of AI work, such as machine learning or facial recognition, as well as the word “AI” itself, so it can spot the infrastructures doing AI-adjacent work, even if AI itself doesn’t come up.

What you can do with it today

See where the work is. Cluster and density views show where humanities-and-AI work concentrates across the UK. Every filter (a tag, an institution, an infrastructure type, the entanglement sliders) reconfigures the whole dashboard at once. The map, the results list, both charts, the word clouds, and both network graphs update as you narrow down your search.

Dial the AI depth. Every record carries an experimental AI entanglement score, from Emerging to Strong, so you can tune the same map from “AI-curious” to “AI-focused.” A separate humanities score works similarly, so you can find infrastructure where the humanities are central or part of the picture. Each infrastructure is scored from its own record (its name, description, and keyword tags) by a fixed rule. The name counts most, the community’s own tags count strongly, and repeated mentions in the description add up. As we have said previously, the score is a thought experiment more than a verdict. It is meant to show how the language an infrastructure uses about itself can shape where it shows up on the map. It is a starting point.

Search in your own words. The map’s vocabulary is donated by its community. Infrastructures tag themselves in their own language, and we have curated those tags into a family tree of fifteen themes that the search understands. Type a niche word and you get your literal matches first; where the view is narrow, the map offers, in one click, to widen it to the theme your word belongs to.

Follow the funding. The Funded Research panel, the first of its kind for our subject-specific maps, sets more than 8,000 funded projects alongside the map, drawn from UKRI’s Gateway to Research with funders from further afield brought in through 360Giving and the NIHR. Filter by funder, council, and project type, sort by date, and see how the funded picture compares with the infrastructure on the map.

Take the data with you. Whatever you are looking at, the filtered infrastructure list or the funded view, downloads as CSV or JSON.

Who is it for?

Different people have different needs and often arrive with different questions. As the map grows, our community’s diversity grows with it. Below you can read some of the map’s use cases.

• A researcher finding their community. Type your own niche word: “ludomusicology” (the study of music in games; the corpus really does have it) returns its one literal match, and the map offers “Broaden to Music Sound and Performance?” to broaden the search to potential collaborators you may otherwise not have come across.
• An early-career researcher looking for opportunities. Filter the Funded Research panel to student and early-career project types and sort by date to see where support is moving, a use our contributors singled out in the sessions. Alternatively, tick the “select only infrastructures that provide funding” from the main filters to leave only the bodies that fund work themselves, a quick shortlist of who might support yours.
• A grant writer scoping a bid. The paired word clouds and the “least associated” view make a hypothesis generator. Where the humanities intersect with AI only weakly could be where a case study, a panel, or a funding bid is waiting. The partner lines in popups and the CSV help with the perennial search for museum and industry partners.
• An institution taking stock. Filter to your own university and every panel becomes that institution’s dashboard; one contributor’s plan was to use it to identify the ways in which the university at large is collaborating with different partners on AI-related initiatives. It’s also a good exercise in case we have missed things – an infrastructure or a partner or even a tag.
• A policy officer or journalist interrogating a slice of AI-related questions. Type “facial recognition” and accept the offer to broaden to Rights and Justice for a consultation-ready view of the civil-society groups and law centres; type “deepfake” and find the synthetic-media work alongside the policy centres studying it.

What our contributors said

Between April and May 2026, before launch, we trialled the prototype with three deliberately small focus groups: more than twenty colleagues, from doctoral researchers to professors and directors, university-based and independent alike. Colleagues called it “an excellent” and “really valuable resource” with one participant remarking that it was “something we’ve really needed for a while.” For researchers who need to know “this whole field,” one of our contributors noted, “your mapping might help me.”

The focus groups changed core components of the map, and they grew it. In February this year, it held just under 200 AI-relevant infrastructures; at launch it holds 316, an increase of around 60 per cent, with the newest arrivals leaning towards the kinds of organisations our contributors asked for. Next week we will tell the story properly: what the focus groups asked for, what we built in response, and how the dataset developed with the needs of the community.

We’re at a crucial moment in the development of AI, and the humanities have a huge role to play in determining our technological future. Simply showcasing what’s happening in humanities research in the UK is incredibly important and helpful to our common cause.

Dr Kathryn Brown, Loughborough University

Try it, then tell us what’s missing

The map is a living resource, representing the centres, institutes, networks, collectives, and societies where this field is taking shape. The breadth of humanities engagement with AI is certainly greater than any dataset can show. That is where you come in.

• Explore Mapping the Humanities and AI, apply a couple of filters, and see whether your work is represented.
• Add your infrastructure if it is missing.
• Tell us what you find, and what we have missed, at mapping@sas.ac.uk.