AI's most consequential questions are humanistic. Who flourishes, and on whose terms? What do we owe one another? How should knowledge be made, shared, and governed? For fifty years, the Integrated Program in Humane Studies has trained students to ask exactly these questions. In 2016 we put them at the center of an AI curriculum — the first program in the world to do so. A decade later that work has been downloaded nearly 100,000 times across 198 countries, and a growing movement of humanities-led AI research is taking shape around the conviction we have always held: that the meaning, purpose, and evaluation of AI are humanistic problems first.

The humanistic frame

Decisions about what AI should do, whom it should serve, how it should be evaluated, and how it should be governed are not technical residue left over once the engineering is finished. They are the questions the engineering exists to answer. Philosophy, ethics, hermeneutics, intellectual history, critical theory, and the literary, artistic, and political traditions that carry them forward are the disciplines that have spent millennia clarifying these questions. Treating them as primary — not as oversight functions, not as compliance — is what humanistic inquiry contributes to AI work.

This is the conviction that founded IPHS in 1975, and the conviction that drove the launch of the world's first human-centered AI curriculum in 2016. The technology has changed; the questions have not.


How humanistic inquiry moves through AI work

Every IPHS research project follows a loop in which humanistic questions enter at the start and humanistic evaluation closes the loop at the end — with AI tooling in the middle, not at the wheel.

01
Humanistic question
02
Method & sources
03
AI tooling
04
Humanistic evaluation
05
Public release

The loop is not a workflow imposed on top of standard practice. It is what practice looks like when meaning, purpose, and evaluation are treated as the work itself.


Three projects, one frame

Each of these projects begins from a humanistic question that long predates AI, then uses contemporary AI tooling to extend the inquiry. The inquiry remains the work; AI is the instrument.

"Who decides what culture survives, and how can endangered communities keep that authority?"
Rescuing Endangered Heritage
Community-governed AI tools for preserving New Orleans jazz archives before they are lost. Sovereignty over the data stays with the community whose heritage it represents.
Schmidt Sciences HAVI
"What does it mean for a system to act ethically, and who is competent to evaluate it?"
Auditing AI as Ethical Inquiry
Principal Investigators in the NIST Center for AI Standards and Innovation. LLM evaluation, red-teaming, and persuasion-manipulation-deception benchmarks read as ethical and political evaluation, not just technical testing.
NIST CAISI
"How does feeling move through a story, and what does measuring it reveal about how we read?"
Reading Literature at Scale
SentimentArcs — the first large-ensemble computational methodology for diachronic sentiment analysis in full-length literary narratives. Open-source. Used worldwide.
ICML 2024 (oral, top 2%)

We are not alone

Humanities-led AI work is growing fast. Major funders — including Schmidt Sciences, the National Endowment for the Humanities, and the Mellon Foundation — are building portfolios that ask humanists, ethicists, and cultural scholars to lead, not advise. Below is a partial list of the cohort IPHS is part of.

Humanistic AI Virtual Institute. 23 awarded teams worldwide bringing humanities, arts, and culture researchers into AI as principal investigators. IPHS is one of the awardees.
Center for AI Standards and Innovation (the U.S. AI Safety Institute Consortium). IPHS faculty represent the Modern Language Association as principal investigators in the humanities track.
An organization of 25,000+ literary and language scholars taking a coordinated role in AI policy, evaluation, and education. IPHS faculty serve in its NIST CAISI delegation.
Funded humanities-led AI research on criminal recidivism, decision-making, and ethical evaluation. IPHS holds two awards in this portfolio.
Convenes interdisciplinary roundtables on AI, consciousness, ethics, and culture — with humanities scholars, artists, and clinicians as conveners.
Alliance of Digital Humanities Organizations — the global federation of DH societies advancing humanities-grounded computational research, including humanities-led AI work.
Oxford Internet Institute lab studying trust, governance, and human-AI interaction with humanities-trained leadership. IPHS collaborates on cross-disciplinary research.
Cultural-policy framework that places culture and humanities at the center of AI's global governance. IPHS faculty have delivered keynote addresses on humanistic AI policy.

This list is partial and growing. If your institution is doing humanities-led AI work and you would like to coordinate, we want to hear from you.


The work needs humanities-trained principal investigators

Humanities-led AI work is not a matter of advisory committees or ethics review boards. It requires principal investigators whose primary expertise is in the humanities, ethics, history, the arts, and culture — with budget authority, intellectual direction, and accountability to the communities the work concerns. Three reasons this matters:

01
Subject-matter judgment
A philosopher of mind, a literary scholar, a historian of religion, or a curator carries decades of training that shapes which questions are worth asking and which answers are credible. That judgment cannot be summarized in a memo.
02
Interpretive method
Hermeneutics, close reading, oral history, archival method, ethnography, philology — these are how meaning is made, contested, and revised. They are how AI outputs become evidence rather than artifacts.
03
Community accountability
Cultural heritage, archival sovereignty, language preservation, ethical risk — these are obligations to specific communities. Humanities researchers carry, and answer to, those obligations as part of professional identity.

Anticipating, not reacting

AI's pace of change is real, and humanities-led AI work has to keep up. IPHS faculty publish on the trajectory of AI progress — open-source generative AI risk (ICML 2024 oral presentation, top 2%), multi-agent and agentic systems, alignment frontiers, and international AI policy (EU, China, U.S.). That work moves the curriculum forward in advance of each new generation of systems, so students are reading what is coming, not what is already deployed.

Anticipation is itself a humanistic skill. It is what intellectual history teaches: that ideas have arcs, that recurrences can be recognized, that the next iteration is rarely a clean break.


Ways to work with us