Grant backs AI for humanities: Mouradian named among recipients
Historian Khatchig Mouradian and his collaborators have been selected for a Schmidt Sciences award to bring AI into core humanities research. Their project, "Connectivity and Individuality in Textual Traditions: Reimagining Scalable and Evidence-Based Approaches for Inclusive and Transformative Humanities Computing," joins a global cohort working under the Humanities and AI Virtual Institute (HAVI).
Schmidt Sciences is allocating $11 million across up to 23 teams to apply AI to fields such as archaeology, history and literature. "Our newest technologies may shed light on our oldest truths, on all that makes us human," said Wendy Schmidt, co-founder of Schmidt Sciences.
Why this matters for researchers
General-purpose AI models favor modern languages, large datasets and flat media. Humanities research deals with sparse, fragmented evidence, ancient or lesser-spoken languages and artifacts that are physical, multi-material and three-dimensional. That mismatch has limited adoption.
HAVI is funding work to build or adapt models that actually meet those constraints-methods that respect context, provenance and cultural nuance while remaining reproducible at scale.
What teams will tackle
- Question answering grounded in specific historical places and periods.
- Film analysis that connects camera movement and soundtracks to narrative structure.
- Studies linking shifts in trade routes or technology to changes in art and literature.
- Discovery of buried archaeological sites using computational approaches.
- Virtual unwrapping of ancient scrolls and reading of torn or shorthand manuscripts.
Work spans industrial England, Qing-era China and ancient Egypt, among other contexts. As HAVI lead Brent Seales notes, "Computational methods have been a part of the study of humanities for decades, and it's time to explore how to integrate AI into this essential scholarship."
Project leadership
- Peter Bol (Harvard University)
- Kianté Brantley (Harvard University)
- Yehuda Halper (Bar Ilan University)
- Unso Jo (Cornell University)
- Khatchig Mouradian (Columbia University)
- Sebastian Nehrdich (Tohoku University)
- Donald Sturgeon (Durham University)
Selection, prior awards and next steps
Teams were chosen after multiple review rounds by Schmidt Sciences and external experts. Two prior HAVI awards this year went to Sorbonne University (artworks of Eugène Delacroix) and EduceLab, a next-gen heritage science facility applying AI and micro-CT to cultural heritage.
The next HAVI application deadline is March 13, 2026. For program details and updates, see Schmidt Sciences.
Practical takeaways for labs and centers
- Plan for low-resource settings: expect small, uneven corpora and ambiguous labels; prioritize approaches like few-shot learning and careful human annotation protocols.
- Work multilingual and diachronic: build pipelines that handle historical orthography, script variation and code-switching.
- Treat provenance as data: track source, context and transformations end-to-end to keep interpretations defensible.
- Think multimodal: combine text, image, 3D scans and material analysis where relevant; align them around clear research questions.
- Co-design with domain experts: pair model development with historians, philologists, conservators and archivists from day one.
- Make results reusable: document datasets, model cards and evaluation protocols so others can replicate and build on the work.
- Address ethics early: engage with communities and stakeholders on cultural sensitivity, data rights and interpretive limits.
Getting started
Audit your collections, define one focused research question and pilot a small, well-documented workflow with human-in-the-loop review. Scale only after the methodology holds up under expert scrutiny.
If your team needs structured upskilling on AI methods and tooling, explore role-based options at Complete AI Training.
Your membership also unlocks:
