Bridging the gap between humanities scholarship and AI capabilities
The Challenge
HAVI aims to address challenges in two core areas: humanities scholarship and AI development.
-
For Humanities Scholarship
Humanistic research is challenging, demanding significant time, energy, and resources to analyze extensive sources for insightful conclusions. However, AI tools present limitations for humanities researchers, as the technology often prioritizes uniformity, often erasing the cultural, material, and perceptual differences that are essential to humanistic research. We believe that combining traditional methods with AI and data-driven approaches can greatly expand access to scholarly resources and enable more rigorous, data-informed research outcomes in the humanities. -
For AI Development
AI models face challenges in multilingual and multimodal contexts common in the humanities. Current AI also struggles with the diverse nature of historical situations, cultural viewpoints, languages, aesthetics, and ambiguities prevalent in humanities scholarship. Conversely, humanities disciplines offer extensive knowledge in areas where AI is weak, such as complex human reasoning, narrative style, metaphorical understanding, and evaluating excellence in ambiguous fields like art and literature. We believe that integrating this humanistic knowledge can significantly advance and improve AI technology.
Program Goals
-
Catalyze Breakthrough Humanities Research Using AI Tools
-
Advance AI Technologies Through Integration Of Humanistic Insights
-
Enable Deep And Equitable Interdisciplinary Collaborations
-
Foster The Development Of A Global And Diverse Research Community
Featured Projects
-
Digital Delacroix at the Sorbonne Center for Artificial Intelligence
-
Artificial Intelligence for Cultural and Historical Reasoning
-
Bridging Large-Scale Computational Analysis and the Close Viewing of Film and Television
-
Augmenting Retrieval for Eurasian Languages
-
Beyond Translation: Opening up the Human Record
-
Meeting the Vesuvius Challenge
-
Playing Heaven: Remapping Early Modern Neo-Confucian Worlds with AI
-
MakingAI: AI-Driven Integration of ‘Messy’ Data in Technical Art History
-
Text Machine: Computing Literary Innovation
-
Discovering Global Archaeological Heritage using AI and Remote Sensing
-
AI Models with Reinforcement Learning to Edit Text in Damaged Manuscripts
-
Musica Subtilior
-
Print & Probability: Using AI to Identify Printers of Clandestine Letterpress Books
-
From Molecules to Masterpieces
-
Medieval Judicial Opinions
-
Seeing Through Style: Multimodal Advances for Iconographic Analysis
-
Decoding the Lost Art of Shorthand
-
An ML toolkit to find hierarchical structure in multi-modal/lingual data
-
Racing to Save History: Can AI Rescue Endangered Archives
-
SETS: A Set-Based Architecture for Knowledge Structures
-
Communities in the Loop: AI for Cultures & Contexts in Multimodal Archives
-
AI for Understanding the Law and its Evolution