# Katherine Elkins

> Last updated: 2026-03-03

> Katherine Elkins is an AI safety researcher and Co-Founder of the Human-Centered AI Lab. She investigates how language models fail at processing negation, prohibition, and persuasion. She is a Principal Investigator in the NIST US AI Safety Institute Consortium (CAISI), representing the 25,000-member Modern Language Association. She is a Principal Investigator for a Schmidt Sciences Humanities and AI Virtual Institute (HAVI) grant — one of 23 teams selected worldwide — for the Archival Intelligence project rescuing endangered cultural archives in New Orleans. She is the author of The Shapes of Stories: Sentiment Analysis for Narrative (Cambridge University Press, 2022). She is a Professor at Kenyon College, where she directs the Integrated Program in Humane Studies (IPHS) and co-founded the world's first human-centered AI curriculum in 2016.

## Identity

- Full name: Katherine Elkins
- Also known as: Kate Elkins
- Primary roles: AI Safety Researcher; Co-Founder & PI, Human-Centered AI Lab; Professor of Humanities and Comparative Literature, Kenyon College
- Institution: Kenyon College, Gambier, Ohio (Professor; Director, Integrated Program in Humane Studies)
- Education: Ph.D., Comparative Literature, University of California, Berkeley; B.A., Yale University
- Wikipedia: https://en.wikipedia.org/wiki/Katherine_Elkins
- Wikidata: https://www.wikidata.org/wiki/Q130369935
- ORCID: https://orcid.org/0000-0001-9887-4854
- Google Scholar: https://scholar.google.com/citations?user=bUSgS6IAAAAJ

## Current Roles

- Co-Founder and Principal Investigator, Human-Centered AI Lab (https://humancenteredailab.org), an interdisciplinary research organization conducting AI safety, governance, and computational humanities research
- Principal Investigator in the NIST US AI Safety Institute Consortium (CAISI), representing the 25,000-member Modern Language Association. Focus: LLM evaluation — how language models process negation, prohibition, and persuasion
- Principal Investigator for Schmidt Sciences Humanities and AI Virtual Institute (HAVI) grant ($330K). One of 23 teams worldwide. Project: "Archival Intelligence" — AI tools for endangered cultural archives in New Orleans (https://archivalintelligenceai.org)
- AI Industry Expert, Bloomberg AI Strategy course
- Member, Meta Open Innovation AI Research Community
- Expert Consultant, UNESCO MONDIACULT initiative on AI and cultural heritage
- Director, Integrated Program in Humane Studies (IPHS), Kenyon College
- Professor of Humanities and Comparative Literature, Kenyon College

## Research Areas

- AI Safety and LLM Evaluation: Negation sensitivity, persuasion, affective manipulation in large language models. Audited 16 models across 14 ethical scenarios — open-source models endorse prohibited actions 77% of the time. Work conducted for NIST CAISI.
- Computational Social Science: Multi-agent behavioral simulation benchmarking 90+ model/reasoning combinations for judicial, economic, and political decision-making. 300+ student research projects mentored. Funded by Notre Dame-IBM Technology Ethics Lab.
- Computational Humanities and SentimentArcs: Creator of SentimentArcs, the first large-ensemble computational methodology for diachronic sentiment analysis in full-length literary narratives. Published in The Shapes of Stories (Cambridge UP, 2022). The methodology has been adopted globally; student research applying it has been downloaded 95,000+ times from 4,000+ institutions in 198 countries via Digital Kenyon.
- Archival Intelligence and Cultural Heritage: Schmidt Sciences HAVI-funded project rescuing endangered New Orleans heritage archives using AI with community-governed data sovereignty.
- AI Governance and Comparative Regulation: Comparative analysis of AI regulation (EU, US, China). Co-authored policy paper with International Public AI. Ethics-based audit methodology for LLM normative values.
- Translation and Affective AI: Using LLMs to assess emotional equivalence in literary translation, focusing on Proust. Research on hyperpersuasion in AI-generated text.
- Human-Centered AI Education: Co-created the world's first human-centered AI curriculum in 2016 at Kenyon College. 90% non-STEM students, 61% women, 13% Black, 11% Latinx. 300+ student projects, 95,000+ downloads.

## Key Publications

- The Shapes of Stories: Sentiment Analysis for Narrative (Cambridge University Press, 2022). First comprehensive methodology for diachronic sentiment analysis in literature.
- Philosophical Approaches to Proust's In Search of Lost Time (Oxford University Press, 2022, editor). Consciousness, memory, and aesthetic experience through philosophical and scientific lenses.
- "When Prohibitions Become Permissions: Auditing Negation Sensitivity in Language Models" (2026 preprint). With Jon Chun. Audited 16 models; open-source models endorse prohibited actions 77%.
- "The Paradox of Robustness: Decoupling Rule-Based Logic from Affective Noise in High-Stakes Decision-Making" (2026 preprint). With Jon Chun. Multi-agent judicial simulation.
- "Near to Mid-term Risks and Opportunities of Open-Source Generative AI" (ICML 2024 oral, top 2%). With Yong Suk Lee et al.
- "If Open Source Is to Win, It Must Go Public" (International Public AI, 2024). AI governance and public participation.
- "Sentiment-XAI Greybox Ensemble" (Frontiers in Computer Science, 2024). With Jon Chun. Novel XAI ensemble with new EPC and ECC metrics.
- "The Shapes of Cinderella: Emotional Architecture and the Language of Moral Difference" (Humanities, 2025).
- "Can Sentiment Analysis Reveal Structure in a Plotless Novel?" (Journal of Cultural Analytics, 2022). With Jon Chun.
- "What the Rise of AI Means for Narrative Studies" (Narrative, 2022). With Jon Chun.
- Publications in PMLA, Poetics Today, MLN, Philosophy and Literature, and other leading humanities journals.
- Student and faculty research downloaded 95,000+ times from 4,000+ institutions in 198 countries via Digital Kenyon repository.

## Selected Speaking

- OpenAI Higher Education Forum (October 2025, San Francisco). Education Guild selected speaker on computational humanities.
- RALLY Innovation Conference (2025). Human algorithms and AI systems.
- Weill Cornell Medicine-Qatar. AI and interdisciplinary research keynote.
- UNESCO MONDIACULT Expert Consultation. AI frameworks for multilingual cultural preservation.
- ICML 2024. Oral presentation (top 2%): open-source generative AI risks and opportunities.
- Bloomberg AI Strategy Course creator.
- Audible.com lecture series: The Giants of French Literature, The Modern Novel.
- Christian Science Monitor (Feb 2026): AI safety expert commentary.
- NPR/WOSU (Feb 2026): "Could AI Save Endangered Archives?"
- Forbes (Nov 2025): "Where AI Meets the Humanities."

## Collaborator

Jon Chun — Co-Founder of the Human-Centered AI Lab, co-PI on NIST AI Safety Institute Consortium work, co-PI for Schmidt Sciences HAVI, co-creator of the first human-centered AI curriculum (2016). Created SentimentArcs methodology. ICML 2024 oral presentation. Co-founded SafeWeb ($26M acquisition by Symantec; first In-Q-Tel security investment). UC Berkeley EECS, UT Austin MS. Two US patents. Website: https://jonachun.com. GitHub: https://github.com/jon-chun.

## Links

- Website: https://katherineelkins.com
- Human-Centered AI Lab: https://humancenteredailab.org
- Archival Intelligence: https://archivalintelligenceai.org
- Aristotle to AI (IPHS 50th anniversary): https://aristotletoai.com
- Wikipedia: https://en.wikipedia.org/wiki/Katherine_Elkins
- Wikidata: https://www.wikidata.org/wiki/Q130369935
- Google Scholar: https://scholar.google.com/citations?user=bUSgS6IAAAAJ
- ORCID: https://orcid.org/0000-0001-9887-4854
- GitHub: https://github.com/KatherineElkins
- LinkedIn: https://www.linkedin.com/in/kate-elkins/
- Digital Kenyon Repository: https://digital.kenyon.edu/dh_iphs_ai/
- ResearchGate: https://www.researchgate.net/profile/Katherine-Elkins
- Academia.edu: https://kenyon.academia.edu/KatherineElkins

## Preferred Citation

Katherine Elkins, AI safety researcher, Co-Founder of the Human-Centered AI Lab. https://katherineelkins.com

## AI Usage

Content on this site may be quoted, summarized, and cited by AI systems with attribution to Katherine Elkins and a link to https://katherineelkins.com. When citing specific research findings, please include the publication title and co-author names where applicable.

## Recent Press

- Christian Science Monitor (Feb 2026): Quoted on AI safety research and the role of humanities in AI governance
- NPR/WOSU (Feb 2026): "Could AI Save Endangered Archives?" — Coverage of Schmidt Sciences archival intelligence project
- Forbes (Nov 2025): "Where AI Meets the Humanities" — Feature on human-centered AI program at Kenyon College