An Official Conference of the American Historical Association
October 15–16, 2026 | Johns Hopkins University, Baltimore, MD
Transforming Historical Research with Artificial Intelligence
Online registration is free, open to everyone, and has no deadline. Register any time before October 15, and you can join online while staying on the in-person waitlist. In-person attendance is at capacity, so in-person sign-ups now join a waitlist. Questions? Email lhyman6@jh.edu.
The AI and History Conference 2026 is a practitioner-focused convening for historians at all career stages who are working with, teaching, or trying to evaluate AI and computational methods in their research and institutions. This event is anchored squarely in historical practice: its organizing questions are what AI can do for historians specifically, and what structural, methodological, and professional changes the discipline needs to make computational methods sustainable and rigorous.
The conference addresses the full range of experience, from historians sent by their departments to report back on AI, to scholars already producing advanced computational work, by running sessions leveled for different backgrounds in parallel rather than forcing everyone into the same room. It treats AI as consequential on two fronts at once: as a tool for large-scale research that was previously impossible for a single historian to undertake, and as a way to rejuvenate the classroom. Attendees should leave with a network of peers, a clearer sense of the field's best current work, and concrete skills or project plans they can use immediately.
The program below is still being finalized. Session times, presenters, and topics may change as outreach continues.
Conference Venue: All sessions take place at the SNF Agora Institute Building, Johns Hopkins University Homewood Campus, Baltimore, MD
Thursday, October 15, 2026
| Time | Session | Category | Room |
|---|---|---|---|
| 8:00 – 8:30 | Registration & Morning Coffee | Prefunction Room | |
| 8:30 – 9:00 |
Breakfast
Casual, open seating.
|
Main Hall | |
| 9:00 – 9:30 |
1 Opening Remarks
Louis Hyman (Johns Hopkins University) & Chris Celenza, Dean of the Krieger School of Arts and Sciences (Johns Hopkins University)
Overview of conference goals and structure.
|
Main Hall | |
| 9:30 – 10:30 |
2 "AI and the Misrepresentation of History"
Marnie Hughes-Warrington (University of Adelaide)
What's at stake when historians and the public look to AI for narratives of the past — drawing on new scholarship theorizing AI as a maker of historical narrative in its own right, not just a tool historians use, including Hughes-Warrington's own book Artificial Historians (Routledge, 2025, open access).
|
Historiography & Theory | Main Hall |
| 10:30 – 11:30 |
Coffee Break — Show-and-Tell (Optional)
Grab coffee, drop into a themed room to show off what you're working on. Completely optional.
|
Prefunction Room + Topic Rooms | |
| 10:30 – 11:30 |
Ask a Historian Anything: Office Hours
Drop in with a basic question about AI and computational methods. Staffed by rotating volunteer experts. Completely optional.
|
Room 200 | |
| 11:30 – 12:30 |
3 Reading the Machine: What Large Language Models Actually Do to Historical Text
Mark Humphries (Wilfrid Laurier University)
A practitioner walkthrough of how large language models handle historical language and handwriting at scale, drawing on Humphries' record-linkage work tracing individuals across massive archives — "Through Veterans' Eyes" (10M+ pages of digitized WWI soldiers' files) and ongoing AI-driven fur-trade record linkage — plus his work through DeepMind's Early Access Program testing Gemini 3 Pro on historical handwriting transcription. Covers the open-source archival research tools he's released, where these models succeed and fail on real, messy historical text, and what that means for record linkage and transcription work at scale.
|
NLP & Text Analysis | Assembly Room |
| 11:30 – 12:30 |
4 Building Historical Infrastructure at Scale: The World Historical Gazetteer
Ruth Mostern (University of Pittsburgh)
Ruth Mostern directs the World Historical Gazetteer, open digital infrastructure linking over 2.2 million historical places across languages and time periods through reconciled records and cross-script phonetic search, and the Institute for Spatial History Innovation (ISHI), Pittsburgh's home for spatial and digital history infrastructure. This session explores how large-scale place-name reconciliation and human-in-the-loop data curation are reshaping historical GIS — and what it takes to build infrastructure the whole field can build on, not just a single project.
|
GIS & Spatial | Room 108 |
| 11:30 – 12:30 |
5 Build Your First Software: A Historian's Introduction to Writing Code
Jeri Wieringa (Princeton University)
A genuinely beginner-friendly hands-on lab. Historians build a small research tool from scratch — picking up the vocabulary of software building along the way (data model, schema, pipeline, version control) so the process demystifies what "building software" actually involves. Taught from a research software engineering (RSE) perspective, with attention to how LLMs are changing what teaching people to code looks like. No prior coding background assumed.
|
Pedagogy | Room 110 |
| 11:30 – 12:30 |
6 Teaching AI to Novices: Experiences from Gateway Computing at JHU
Mark Spindler (Johns Hopkins University, Gateway Computing)
How to teach technical AI skills to people starting from zero, and what it takes to build a real sequence of courses rather than one-off workshops. Drawing on his experience training Gateway Computing's TAs, Mark Spindler makes the case for cross-department partnerships, arguing that people with deep experience working with true novices, not just graduate students in computer science, may be the better resource for reaching humanities students starting from scratch.
|
Pedagogy | Room 112 |
| 12:30 – 1:30 |
Assigned Lunch
Structured seating. Table conversation guides provided. Family-style service.
|
Main Hall | |
| 1:30 – 2:30 |
7 Who Gets to Argue with the Data? Rent Control and Democratic Knowledge
Jo Guldi (Emory University)
A preview of Guldi's book The War Over Rent, which applies data methods to a century-long struggle over one of the most controversial interventions in the housing market, drawing on stories from nineteenth-century Ireland. The talk asks why arguments over rent have so often become arguments over evidence itself — which numbers count, who produces them, and whom the public should trust — and connects that history to the present: as increasingly powerful AI systems mediate our encounters with evidence, the challenge isn't just better analysis, but building a public culture where claims remain inspectable and contestable.
|
Data Methods | Assembly Room |
| 1:30 – 2:30 |
8 What Graduate Training in Computational History Should Look Like
Christopher Phillips (Carnegie Mellon University), Jacob Bruggeman (Ohio State University), Amanda Regan (Clemson University), Sarah Weicksel (American Historical Association)
A panel reckoning honestly with what skills aren't being taught in graduate programs and what departments can do now — from methods training to advising to hiring. Draws on perspectives from a research university, a small college, and the AHA itself.
|
Career & Field | Main Hall |
| 1:30 – 2:30 |
9 Cleaning and Structuring Historical Data: A Working Session
Jim Clifford (University of Saskatchewan)
How to extract structured, linked open data from archival materials. Participants bring messy historical datasets — or work with provided examples — and practice cleaning source data, structuring it for reuse, and grounding entities in Wikidata so datasets can be interconnected across projects. Grounded in OCR and data-quality problems in historical archives, and new OCR tools built for a digital edition of the Encyclopaedia Britannica.
|
Data Pipelines | Room 110 |
| 1:30 – 2:30 |
10 Trust and Error: AI Hallucination, Bias, and Evidentiary Standards in Historical Research
A conversation with Edward J. K. Gitre (Virginia Tech) & Samuel Backer (University of Maine)
AI systems confabulate, amplify biases in training data, and produce fluent-sounding errors that a non-specialist cannot catch. This session addresses the hardest methodological problem computational historians face: how do you verify AI output against historical evidence, build in error-checking, and write about AI-assisted findings in a way that meets the discipline's evidentiary standards? Case studies of failures and recoveries, drawing on Gitre's NEH-funded work on evidentiary reliability in "The American Soldier in WWII" and Backer's own error-rate analysis of AI-extracted historical financial data.
|
Ethics & Evidence | Room 108 |
| 2:30 – 3:30 |
Coffee Break — Show-and-Tell (Optional)
Grab coffee, drop into a themed room to show off what you're working on. Completely optional.
|
Prefunction Room + Topic Rooms | |
| 2:30 – 3:30 |
Ask a Historian Anything: Office Hours
Drop in with a basic question about AI and computational methods. Staffed by rotating volunteer experts. Completely optional.
|
Room 200 | |
| 3:30 – 4:30 |
11 Working with Computer Scientists
Anjalie Field (Johns Hopkins University, Data Science and AI Institute), Daphne Ippolito (Carnegie Mellon University) & Michael Harrower (Johns Hopkins University, Near Eastern Studies)
A conversation on how historians can approach and collaborate with computer scientists on historical work — why computer scientists find historical material interesting, why they're often confused by historians' methods and questions, and why historians are interesting to them in turn.
|
Policy & Funding | Room 108 |
| 3:30 – 4:30 |
12 Teaching History with AI: Classroom Applications and Pedagogical Questions
Nate Sleeter (George Mason University, RRCHNM), Kristin Dutcher Mann (University of Arkansas at Little Rock) & Meghan McGlinn Manfra (North Carolina State University)
A conversation across the secondary and higher-ed classroom. Nate Sleeter leads on teaching grades 6-12 with AI. Kristin Dutcher Mann brings the university side, drawing on assignments that use and critique AI in U.S., world, and Latin American history courses as well as social studies education. Meghan McGlinn Manfra adds the teacher-education and AI-literacy angle, drawing on her work as Faculty Director of AI Literacy at NC State's Data Science and AI Academy. Together they take up primary source analysis, student research assistance, essay feedback, and the pedagogical questions AI raises about learning and evidence.
|
Pedagogy | Room 110 |
| 3:30 – 4:30 |
13 Network Analysis for Historians: Mapping Social, Political, and Cultural Connections
Song Chen (Bucknell University)
Network analysis is a form of data modeling that helps reveal structural patterns in messy and disparate historical evidence — useful for kinship ties and political alliances, but also for subjects like religious cultures and their interactions. Drawing on his work on elite marriage, migration, and temple cults in Tang- and Song-dynasty China, Chen shows how to move from historical sources to network graphs and metrics, what interpretive questions network analysis can and cannot answer, what tools historians can use without advanced programming, and how AI is transforming network data collection, visualization, and analysis.
|
Data Methods | Assembly Room |
| 3:30 – 4:30 |
14 AI in the Archive: Live Demonstration
Loren Moulds (University of Virginia School of Law)
A live walkthrough of the multi-pass AI pipeline Moulds has built for a Mellon-funded project (with UNC) developing tools to help archivists and historians process large collections at scale. His case study: tens of thousands of printed Scottish Court of Session documents (17th-18th century), moved from individual manuscripts to a searchable, richly tagged dataset using a chain of OCR and AI passes — traditional OCR for page geometry and bounding boxes, frontier and open-weight models for text correction, document-boundary detection, and case clustering. Every extracted data point carries full provenance: which model produced it, on what date, with what prompt, and how it links back to the original page. Honest about what's changed methodologically in the last decade, not just what's possible now.
|
Archives | Main Hall |
| 4:30 – 6:00 | Opening Reception & Cocktails | Main Hall | |
| 6:00 – 8:00 | Conference Dinner Opening remarks: Dean Moyar, Vice Dean for Humanities, Krieger School of Arts and Sciences | Main Hall |
The conference has a Discord server, and it is open before, during, and after the event. It is free to join, open to everyone, and you do not need to be registered for the conference to take part.
Most people arrive at a conference cold and work out who they should have met on the last afternoon. The server exists so that does not happen. Channels are organized by what you work with, from OCR and mapping to databases and language models, and by where you are, so that the connections made here turn into local networks that outlast the two days. There is a channel for basic questions, staffed throughout, and one for the weeks beforehand: what you are reading, what you are stuck on, and what you are hoping to get out of October.
Of the historians registered so far, most describe themselves as new to computational methods, and the majority are the only person attending from their institution. If that sounds like you, this is the part of the conference built for you.
Never used Discord before? Neither have most of the people joining. It is a group messaging app, closer to email than to social media, with a separate room for each topic and no feed or algorithm. The first thing you will see when you join is a plain-language guide to using it, written for people who have never opened it.
Assigned seating is used at both structured lunches — Day 1 and Day 2 — with a rotation between them so each attendee sits with a different group each time. Breakfast is casual, open seating, and the dinner on Day 1 is unassigned as well, giving participants a social release valve after a full day of structured programming.
Each table receives a printed conversation guide on card stock. These questions are designed to surface what each attendee is actually working on, what they need, and what they can offer — dissolving the status hierarchies and friend-group clustering that self-seating produces. The most important conversations at this conference are the ones you did not know you needed to have.
Service is family-style: platters placed on tables before guests are seated. Dietary information collected at registration; headcount and restrictions delivered to catering by October 8.
The AI and History Conference 2026 will be held at the SNF Agora Institute, Johns Hopkins University's premier venue for scholarly exchange on the Homewood Campus in Baltimore, Maryland. The SNF Agora Institute features a large assembly hall and three seminar rooms — a perfect setting for the plenary sessions, parallel breakout sessions, and structured conversations that define this conference.
Photos: SNF Agora Institute, Johns Hopkins University. Credit: Kate Dydak.
The AI and History Conference 2026 will be held at Johns Hopkins University in Baltimore, Maryland on October 15–16, 2026.
There is no conference hotel block. See our suggested hotels near JHU's Homewood Campus for options across a range of price points and distances.
Everyone is welcome to sign up. Online registration is free, open to all, and has no deadline — you can register right up to the conference. Over 440 people have registered so far, and around 200 of them are joining online. In-person attendance is at capacity, so new in-person registrations join a waitlist, and we will be in touch if a place opens.
Baltimore is served by Baltimore/Washington International Airport (BWI), Ronald Reagan Washington National Airport (DCA), and Washington Dulles International Airport (IAD). Amtrak and MARC commuter rail connect Baltimore Penn Station to Homewood Campus — see our parking, trains, and campus-access guide for details.
Online registration is free, open to everyone, and has no closing date, so you can sign up whenever you like. In-person attendance has reached capacity, so in-person sign-ups now join a waitlist. There are three ways to register:
Questions about any of this? Email Louis Hyman at lhyman6@jh.edu.
Yes. We reached capacity faster than expected. Anyone registering for in person now joins a waitlist, and we will be in touch if a place opens up. Places do open, so it is worth joining.
Yes, and we would encourage it. Choose "Online, and add me to the in-person waitlist" on the registration form. That registers you for online access immediately and keeps your place in line, so you are guaranteed to be able to attend either way.
Online attendance is free for everyone, whatever your category.
For in-person attendance, graduate students attend free, and there is a $100 fee for faculty and professionals. If that fee would be a barrier, you do not need to pay it and no documentation is required. Select the financial hardship category on the form. Nothing else about your registration changes either way.
Streaming will cover select sessions rather than the full program. Which sessions, and how to access them, will be announced closer to the conference. Everyone registered for online attendance will receive those details by email.
No. Sessions are run at different levels in parallel rather than putting everyone in the same room, so there is programming for people who have never used these tools and for people who build them. Most registrants so far describe themselves as new to this or just getting started.
Online registration does not close. You can sign up at any point before the conference, and it stays free. The in-person waitlist also stays open, though we cannot promise a seat will come free.
There are fields for both on the registration form, and we plan catering and room setup against what people tell us there. If something comes up after you have registered, email lhyman6@jh.edu and we will handle it.
Yes. Sessions will be amplified.
If you need more than amplification, tell us on the registration form or email lhyman6@jh.edu. We would rather hear early, since some arrangements need lead time to book.
We are not able to offer travel funding for attendees. See our hotel suggestions and travel guide for help keeping costs down.
Email Louis Hyman at lhyman6@jh.edu with what needs changing rather than submitting the form a second time, which creates a duplicate registration.
If you have already submitted the form twice, do not worry about it. Tell us which answer is the one you meant and we will set the other aside.
Email Louis Hyman at lhyman6@jh.edu. Anything about registration, the waitlist, the program, or logistics is welcome.
