Professor of Humanities & Comparative Literature · Faculty in Computing · Kenyon College

Katherine
Elkins

Researcher and educator whose work bridges humanities and social sciences with empirical AI research, security evaluation, and policy development. Among the first scholars to fine-tune and systematically evaluate GPT-2 (2019), with continuous hands-on technical implementation through contemporary LLMs. Co-creator of the first human-centered AI curriculum, integrating computational methods with ethics, governance, and humanistic inquiry since 2016.

Katherine Elkins
Federal AI Safety
Principal Investigator
US AI Safety Institute (NIST CAISI)
Schmidt Sciences
PI, HAVI Award
1 of 23 worldwide · Archival Intelligence
Cambridge University Press
The Shapes of Stories
First methodology for emotional arc in narrative
OpenAI · Bloomberg · UNESCO
Keynote Speaker
Higher Ed Forum · AI Strategy · MONDIACULT Cairo
95k+
Downloads of mentored student research from 4,000+ institutions worldwide
2016
Co-created the world's first human-centered AI curriculum
17+
Peer-reviewed publications across AI, humanities, and governance
61%
Women enrolled in AI courses · 13% Black · 11% Latinx
Research

Where AI meets human systems

AI Safety & Security

Red-teaming LLMs, syntactic framing vulnerabilities, threat modeling for multi-agent architectures. Developing evaluation frameworks for autonomous AI systems at NIST CAISI.

Governance & Regulation

Comparative global AI regulation (EU, China, US). Open-source policy. UNESCO cultural heritage frameworks. Behavioral prediction and ethical auditing of AI decision-making systems.

Computational Humanities & Social Sciences

Sentiment analysis for narrative, affective AI, multi-agent simulation of high-stakes human decisions. Created the SentimentArcs methodology adopted by researchers globally across literary, medical, and social science applications.

Archival Intelligence

Schmidt Sciences HAVI project rescuing endangered New Orleans heritage. Community-governed data sovereignty, privacy protections for historically marginalized populations.

Computational Methods & Security
"AgenticSimLaw: A Juvenile Courtroom Multi-Agent Debate Simulation for Explainable High-Stakes Tabular Decision Making"
Accepted, LaMAS 2026 (with J. Chun and Y. S. Lee, Notre Dame) · Multi-agent judicial simulation
"Syntactic Framing Fragility: An Audit of Robustness in LLM Ethical Decisions"
arXiv, December 2025 (with J. Chun)
"Beyond Plot: How Sentiment Analysis Reshapes Our Understanding of Narrative Structure"
Journal of Cultural Analytics, September 2025
"Can GPT-3 Pass a Writer's Turing Test?"
Journal of Cultural Analytics 5(2), 2020 (with J. Chun) · 362+ citations
Governance, Security & Policy
"Risks and Opportunities of Open-Source Generative AI"
Proceedings of ICML 2024 · Oral presentation (top 2%)
"If Open Source is to Win it Must Go Public"
Spotlight talk, CODEML, ICML 2025
"Informed AI Regulation: Comparing the Ethical Frameworks of Leading LLM Chatbots"
arXiv, 2024 (with J. Chun)
"Comparative Global AI Regulation: EU, China, and the US"
arXiv, 2024 (with C. Schroeder de Witt & J. Chun)
Interdisciplinary Methods & Theory
"A(I) University in Ruins: What Remains in a World with Large Language Models?"
PMLA, August 2024
"AI Comes for the Author"
Poetics Today 45(2), 2024
"What the Rise of AI Means for Narrative Studies"
Narrative 30(1), 2022 (with J. Chun)

Additional papers under review at ICML, FAccT, and other venues. Forthcoming in Modern Fiction Studies, Narrative Inquiry, and UNESCO. Full list on Google Scholar →

Books

Published works

The Shapes of Stories book cover
Cambridge University Press · 2022
The Shapes of Stories: Sentiment Analysis for Narrative
Developed the first robust methodology for surfacing emotional arcs in stories, comparing over three dozen models including the latest deep learning AI. Part of the Elements in Digital Literary Studies series.
Proust's In Search of Lost Time book cover
Oxford University Press · 2022
Proust's In Search of Lost Time: Philosophical Perspectives
Editor and contributor. Eight essays from prominent philosophers and critics treating temporality, music, love, consciousness, and fiction in one of the longest and most complex novels ever written.
Audible / Modern Scholar · 2009–2013
Lecture Series
Three audiobook lecture series with an international audience: The Modern Novel; The Giants of French Literature; Odyssey of the West.
Speaking & Keynotes

From Cairo to Yale

Katherine Elkins presenting Shapes of Stories at OpenAI Higher Education Forum Katherine Elkins speaking at RALLY Innovation 2025
OpenAI Higher Education Forum
October 2025 Invited
Weill Cornell Medicine-Qatar, Doha
October 2025 · CME credits Keynote
UNESCO MONDIACULT, Cairo
April 2025 · Global AI policy framework Invited
Concordia College — Opening Plenary
Faith, Reason & World Affairs Symposium, September 2025
Kahn Liberal Arts Institute, Smith College
February 2025
Northrup Distinguished Lecture, Yale
"Can AI Reason Ethically?" · February 2024 Distinguished
RALLY Innovation
"Human Algorithms" · 2025
Carleton College Day of Digital Humanities
May 2024
Deloitte HumanCentric Labs
"How AI is Defining the Future of Organizations" · 2024
US AI Safety Institute / NIST CAISI
"Emotional Hacking of LLMs" · December 2024
International Society for the Study of Narrative
2025 & 2026
Worcester Polytechnic Institute
"AI at a Crossroads: Who Decides Our Future?" · February 2025
Policy & Governance

Shaping how AI is governed

NIST CAISI
Principal Investigator
US AI Safety Institute. Developing evaluation frameworks for autonomous AI systems. Red-teaming, threat modeling, syntactic vulnerability detection.
Schmidt Sciences
Principal Investigator
HAVI Award (1 of 23 worldwide). "Archival Intelligence: Rescuing New Orleans Heritage." Data sovereignty, community consent, privacy protections.
UNESCO
AI & Cultural Heritage Initiative
Contributing to 2025 global policy framework. MONDIACULT Cairo. Forthcoming article on AI's impact on cultural preservation.
Notre Dame–IBM
Co-PI, Tech Ethics Lab
"How Well Can GenAI Predict Human Behavior?" Multi-agent framework testing autonomous system behavior in high-stakes judicial and social contexts. 90+ model/reasoning combinations benchmarked.
Public AI
Member
International network advancing public infrastructure for AI. Co-author of ICML 2025 spotlight talk on open-source policy.
OpenAI
Higher Education Forum
Selected speaker at inaugural forum. Addressing AI integration in academic research and curriculum.
Featured In

In the press

Forbes Bloomberg Al Jazeera Chronicle of Higher Ed NPR / WOSU
"Could Artificial Intelligence Save Endangered Archives?" — NPR / WOSU Public Media, February 2026
"Where AI Meets The Humanities: Inside Kenyon College's Bold Experiment" — Forbes, November 2025
AI Strategy Course — Bloomberg, on-screen presenter and industry expert, 2024
"Is AI Better at Making Art Than Humans?" — Al Jazeera debate, April 2023
"Developing AI Programs" — Chronicle of Higher Education Virtual Forum, June 2025
Get in Touch

Let's connect

info@katherineelkins.com
Google Scholar LinkedIn Wikipedia