# Jon A. Chun — Full Site Content
# Source: https://jonachun.com
# Last updated: 2026-09-06
# This is the full-text companion to llms.txt (summary version)

## Homepage (/)

Jon A. Chun is an AI evaluation researcher and builder working on model reasoning, human judgment, agent behavior, computational methods, and research infrastructure.

He serves as Co-PI representing the Modern Language Association in the NIST CAISI consortium and as Co-PI on Archival Intelligence, part of Schmidt Sciences' Humanities and AI Virtual Institute, developing computational methods while rescuing endangered cultural archives. At Kenyon, Jon designed the technical curriculum and courses and worked with Katherine Elkins to develop the broader interdisciplinary educational model that became the world's first Human-Centered AI curriculum and lab in 2016. Earlier, he co-founded and led SafeWeb.

His work connects four areas: frontier-model evaluation; computational methods for difficult human phenomena; deployed technical systems and research infrastructure; and technical capability building that moves students and collaborators toward original research.

## Human-Centered AI (/human-centered-ai)

Human-Centered AI is a technical teaching and research environment, not a general AI-literacy program. At Kenyon College it enables humanities and social-science students to formulate research questions, build and evaluate computational systems, examine their limitations, and publish original work. Jon designed the technical curriculum, computing courses, research workflows, and progression from first programming work to independent computational research; Jon and Katherine Elkins developed the broader interdisciplinary educational model together. The curriculum and lab began in 2016. Program: https://humanestudies.org

## Research (/research)

Frontier-model evaluation: as Co-PI representing the Modern Language Association in the NIST CAISI consortium, Jon works on LLM evaluation, red-teaming, and ethical auditing. Chun and Elkins conducted the first ethics-based audit of moral reasoning in deployed LLMs; the team's results were presented during the opening keynote at the consortium's first plenary at the University of Maryland. Governance is a supporting thread: with collaborators at Oxford and elsewhere he co-authored the first systematic comparison of EU, China, and US AI regulation after passage of the EU AI Act (doi:10.48550/arXiv.2410.21279), and was among the authors of the open-source generative-AI position paper accepted as an ICML 2024 oral presentation.

Computational narrative and digital humanities: Jon's work includes the first rigorous methodology for narrative sentiment analysis (2019), the contribution of "middle reading," the first writer's Turing test of an LLM (2020), SentimentArcs' novel ensemble method (2021), the first application of XAI to narrative (2023), and the first multimodal method measuring long-form film sentiment-arc coherence across modalities (2024). Elkins's complementary program examines what stays the same and what changes when stories travel across languages, cultures, and time.

What the work established, and when:
- 2019 — Narrative sentiment and "middle reading." "Can Sentiment Analysis Reveal Structure in a Plotless Novel?" established the first rigorous methodology for narrative sentiment analysis and named a third position between distant and close reading. Earlier studies had applied sentiment tools to narrative; this work supplied the rigorous comparative methodology.
- 2021 — Ensembles, not models. SentimentArcs (arXiv:2110.09454) established that diachronic sentiment analysis of narrative needs an ensemble rather than a model: dozens of models weighed against one another, with smoothing and model selection treated as interpretive choices rather than neutral preprocessing. It brought dynamic time warping into diachronic sentiment analysis, using DTW to compute the distance between two whole arcs while absorbing the temporal shifts and stretches that separate otherwise similar story shapes; paired with LTTB downsampling, which normalizes every arc to a common number of points while preserving peaks, valleys, and endpoints, those DTW distances drive hierarchical clustering — which is what makes arcs of unequal length comparable at all. It also reported the negative result that state-of-the-art transformers can struggle to find narrative arcs — a limit of scale, not of tuning. The open-source implementation (maintained since 2019) is the technical foundation of The Shapes of Stories (Cambridge University Press, 2022).
- 2023 — Explainability and comparability. "eXplainable AI with GPT-4 for Story Analysis and Generation" brought XAI to narrative, where the technique had been largely confined to image classifiers, and supplied the sentence-level story-trajectory methodology together with the agglomeration and comparability solutions that let narratives of unequal length be compared on shared coordinates.
- 2024 — Across modalities. MultiSentimentArcs made cross-modal coherence measurable: whether the arc a film's dialogue traces and the arc its images trace are in fact the same arc. It is a multimodal method (text and image, not multiple languages).

First AI presentations in the humanities: the International Conference on Narrative (New Orleans, March 2020) and the Modern Language Association convention (Philadelphia, January 2024).

Research threads:
- NIST CAISI evaluation: LLM evaluation and red-teaming for standards-facing work, with ethical auditing as the through-line.
- Comparative AI regulation: mapping policy approaches across the EU, China, and the US with Christian Schroeder de Witt and Katherine Elkins.
- Computational narrative: from the GPT-3 Writer's Turing Test article to narrative theory, sentiment arcs, and AI-assisted interpretation.
- Open-source generative AI: risks and opportunities of open-source generative AI, including the ICML 2024 position paper.
- Archival Intelligence: part of Schmidt Sciences' Humanities and AI Virtual Institute; Jon serves as Co-PI, developing computational methods for cultural archives while rescuing endangered multilingual newspapers and early jazz artifacts.

Selected publications:
- Elkins, K.; Chun, J. (2020). "Can GPT-3 Pass a Writer's Turing Test?" Journal of Cultural Analytics. doi:10.22148/001c.17212
- Chun, J. (2021). "SentimentArcs: A Novel Method for Self-Supervised Sentiment Analysis of Time Series Shows SOTA Transformers Can Struggle Finding Narrative Arcs." arXiv:2110.09454
- Chun, J.; Elkins, K. (2022). "What the Rise of AI Means for Narrative Studies: A Response to 'Why Computers Will Never Read (or Write) Literature' by Angus Fletcher." Narrative 30(1), 104-113. doi:10.1353/nar.2022.0005
- Chun, J.; Elkins, K. (2023). "The Crisis of Artificial Intelligence: A New Digital Humanities Curriculum for Human-Centred AI." IJHAC 17(2). doi:10.3366/ijhac.2023.0310
- Chun, J.; Elkins, K. (2023). "eXplainable AI with GPT4 for story analysis and generation: A novel framework for diachronic sentiment analysis." Int. J. Digital Humanities 5(2-3), 507-532. doi:10.1007/s42803-023-00069-8
- Chun, J. (2024). "MultiSentimentArcs: a novel method to measure coherence in multimodal sentiment analysis for long-form narratives in film." Frontiers in Computer Science 6. doi:10.3389/fcomp.2024.1444549
- Eiras, F.; et al. incl. Chun, J., Elkins, K. (2024). "Position: Near to Mid-term Risks and Opportunities of Open-Source Generative AI." ICML (oral).
- Chun, J.; Schroeder de Witt, C.; Elkins, K. (2024). "Comparative Global AI Regulation: Policy Perspectives from the EU, China, and the US." doi:10.48550/arXiv.2410.21279

Related infrastructure:
- SentimentArcs (github.com/jon-chun/sentiment_arcs): the open-source ensemble, maintained since 2019; the technical foundation of The Shapes of Stories (Cambridge University Press, 2022). Applied to novels, fan fiction, games, film and television scripts, end-of-life medical narratives, and economic-crisis discourse.
- AI-LIT: an open repository for literary AI workflows, used by Katherine Elkins for text processing and visualization in "In Search of a Translator" (Frontiers in Computer Science 6:1444021, 13 August 2024), which compares a literary original with multiple translations across whole-narrative time. doi:10.3389/fcomp.2024.1444021

Reception:
- The GPT-3 creative-writing paper was cited by Floridi and Chiriatti in Minds and Machines, by Spitale and co-authors in Science Advances, and by Mei and co-authors in PNAS.
- The Narrative article on AI and narrative studies was later discussed by James Phelan in Poetics Today.
- The human-centered AI curriculum article was cited by Jaramillo and Chiappe in Prospects.
- The ICML open-source generative-AI position paper was cited by Taeihagh in Policy and Society.
- The comparative AI regulation paper was cited by Floridi and Ascani in Minds and Machines, by Perboli and co-authors in Economic and Political Studies, and by Olugbade in Global Public Policy and Governance.

## Building (/building)

SafeWeb: Jon co-founded SafeWeb with Stephen Hsu and James Hormuzdiar and later served as CEO. SafeWeb ran the world's largest privacy, anonymity, and anti-censorship web proxy, built Triangle Boy, and was In-Q-Tel's first security investment. Its technology was among the world's first browser-based VPN appliances. Symantec acquired SafeWeb for $26 million; Jon then served as director of development for Symantec's clientless VPN appliance line and led the team that launched the Symantec Clientless VPN Gateway 4400 series.

Historical source record: the strongest SafeWeb account includes both the successful anti-censorship / enterprise-security arc and the documented weakness in the consumer anonymizer. The Wall Street Journal covered the In-Q-Tel licensing and investment agreement; The New York Times reported on SafeWeb proxy technology and Chinese internet censorship; RAND analyzed SafeWeb and Triangle Boy in Chinese dissident internet use; Wired, Computerworld, and USENIX Security described JavaScript- and cookie-based vulnerabilities that could deanonymize users; Le Monde, CRN, Network World, and The Register add international and trade-press context around the In-Q-Tel link, SSL VPN shift, and Symantec acquisition.

Key sources:
- The Wall Street Journal: https://www.wsj.com/articles/SB981939629132013437
- The New York Times: https://www.nytimes.com/2001/08/30/technology/us-may-help-chinese-evade-net-censorship.html
- RAND MR-1543: https://www.rand.org/pubs/monograph_reports/MR1543.html
- Wired, SafeWeb vulnerabilities: https://www.wired.com/2002/02/safewebs-holes-contradict-claims
- Wired, SafeWeb response: https://www.wired.com/2002/02/safeweb-promises-security-fix
- USENIX Security: https://www.usenix.org/events/sec02/full_papers/martin/martin.pdf
- CRN acquisition coverage: https://www.crn.com/news/security/18840041/symantec-to-acquire-safeweb
- Network World acquisition coverage: https://www.networkworld.com/article/2337812/symantec-purchases-ssl-vpn-maker-safeweb.html
- The Register acquisition coverage: https://www.theregister.com/2003/10/21/symantec_snaffles_safeweb/

Patents — among the first SSL/clientless VPN appliances:
- US 7,730,528 — "Intelligent secure data manipulation apparatus and method". Originally assigned to SafeWeb, Inc., then to Symantec following the acquisition.
- US 8,065,520 — "Method and apparatus for encrypted communications to a secure server". Assigned to Symantec.

Human-Centered AI Lab: Jon co-founded the nonprofit Human-Centered AI Lab, an umbrella that lets distributed teams of researchers and domain experts secure funding and collaborate on AI across institutional and disciplinary boundaries. https://humancenteredailab.org

## Teaching (/teaching)

Jon and Katherine Elkins co-founded the AI CoLab and the world's first Human-Centered AI curriculum and lab at Kenyon in 2016. Students bring the integrated humane-studies tradition to AI from the inside, building and testing the systems they study.

Courses:
- IPHS 200 — Programming Humanity
- IPHS 300 — AI for Humanity
- IPHS 391 — Frontiers of AI: GenAI Multi-Agent Networks: one of the first interdisciplinary, project-based agentic-AI courses, open to every division of the liberal arts. Approved 2023; first taught Fall 2024; last taught Fall 2025; more than half the content is rewritten each year. Materials: https://github.com/jon-chun/frontiers-of-ai-automating-intelligence-with-multi-agent-frameworks and https://github.com/jon-chun/GenAI-Multi-Agent-Networks-and-Digital-Twins
- IPHS 484 — Senior Seminar

Reach: Jon and Katherine Elkins have mentored 190+ undergraduate research projects published on Digital Kenyon, where they have been downloaded 130,000+ times from 4,700+ institutions across 198 countries. Adopting institutions include Stanford, MIT, Berkeley, Carnegie Mellon, and Princeton.

Co-founder Katherine Elkins has presented the program widely, including at Carleton College's Day of Digital Humanities, the Kahn Institute at Smith College, Washington University, Weill Cornell Medicine–Qatar, and the OpenAI Forum.

## Collaborations (/collaborations)

The work runs across silos that rarely talk to each other.
- Frontier AI labs: OpenAI, Meta (Open Innovation AI Research Community, since 2023), IBM (Notre Dame–IBM Technology Ethics Lab).
- Industry, analysts, integrators, investors, VCs: BWG Global — multi-year participation (since 2023) in live forums across the full AI-market agenda (public LLMs, OpenAI, DeepSeek, Anthropic, Gemini, agentic adoption, benchmarking, enterprise deployment, regulation). Jon Chun: "Participating in BWG forums since 2023 has informed my interdisciplinary work by providing candid, peer-level conversations with active decision-makers about real enterprise AI adoption." (bwgglobal.com)
- Government and standards: NIST CAISI — Co-PI representing the Modern Language Association in the NIST CAISI consortium. The team's ethics-based audit results were presented during the opening keynote at the consortium's first plenary at the University of Maryland.
- Multilateral: UNESCO, UN.
- Nonprofit: The Helix Center; Human-Centered AI Lab.

Archival Intelligence is part of Schmidt Sciences' Humanities and AI Virtual Institute. As Co-PI, Jon develops computational methods, evaluation workflows, and open research infrastructure for cultural archives while preserving their interpretive questions, historical context, and domain expertise, including work rescuing endangered New Orleans Creole and Cajun multilingual newspapers and early jazz artifacts.

Notre Dame–IBM Technology Ethics Lab: $60,000 grant, 1 of 11 internationally, to audit LLM fairness, accuracy, transparency, and explainability in high-stakes decisions with Notre Dame's Yong Suk Lee.

## Writing (/writing)

The Writing page lists selected public essays and posts on AI research, education, governance, programming, and open source, with concise summaries and links to the original publications.

Public writing digest:
- "AI Skills Demand vs CS Unemployment" — contrasts weak entry-level CS hiring with rising demand for AI-fluent workers who combine technical judgment and domain expertise.
- "IPHS at 50: A Decade of Human-Centered AI at Kenyon College" — connects Kenyon's interdisciplinary humane-studies tradition to a decade of human-centered AI teaching.
- "Coding's Crisis" — argues that AI shifts programming value from routine code production toward problem framing, systems judgment, and applied expertise.
- "AI's Shift from the Narrow Technical to the Universal Human" — frames AI as a broad human and institutional challenge, not a narrow technical specialty.
- "AI is Not the New Oil" — pushes back on simple resource metaphors and emphasizes compute, infrastructure, talent, regulation, and geopolitical leverage.
- "The Future of AI is Wide Open (Source)" — considers open-source AI as a force for diffusion, experimentation, governance tension, and competition with closed frontier systems.
- "A First for Higher Education to Meet the Challenge of AI" — argues that higher education needs integrated AI learning across the liberal arts.
- "Playing Catchup in the AI Race" — looks at AI race dynamics and how institutions respond after falling behind.

## Press (/press)

- Forbes (Nov 2025) — "Where AI Meets the Humanities: Inside Kenyon College's Bold Experiment."
- NPR / WOSU (Feb 2026) — "Could AI Save Endangered Archives?"
- Christian Science Monitor (Feb 2026) — on AI safety and the NIST consortium.
- Al Jazeera, The Stream (2023) — debate on AI and art.
- Chronicle of Higher Education (Jun 2025) — virtual forum, invited speaker.

## About (/about)

Jon Chun is an AI researcher, builder, and educator. At Kenyon College he is a visiting instructor of humanities and affiliated scholar in scientific computing. Jon designed the technical curriculum and courses for Kenyon's Human-Centered AI program and worked with Katherine Elkins to develop the broader interdisciplinary educational model; together they established the world's first Human-Centered AI curriculum and lab in 2016. He serves as Co-PI representing the Modern Language Association in the NIST CAISI consortium and as Co-PI on Archival Intelligence, part of Schmidt Sciences' Humanities and AI Virtual Institute. Before Kenyon he co-founded and led SafeWeb and holds two US patents on early VPN-appliance technology. He holds a B.S. in EECS from UC Berkeley and an M.S. from UT Austin.

## Contact (/contact)

For research, collaboration, speaking, media, and professional inquiries, use https://jonachun.com/contact.

Privacy policy (/contact#privacy): a static site on Netlify. No cookies, no client-side storage, no analytics or tracking scripts, no third-party services; fonts are self-hosted. No personal data is collected, stored, or shared.

## Ecosystem

- Humane Studies (IPHS, Kenyon): https://humanestudies.org
- Human-Centered AI Lab: https://humancenteredailab.org
- Katherine Elkins (co-founder): https://katherineelkins.com
- Archival Intelligence: https://archivalintelligenceai.org

## Profiles

- Google Scholar: https://scholar.google.com/citations?user=l-iUHQMAAAAJ&hl=en
- GitHub: https://github.com/jon-chun
- LinkedIn: https://www.linkedin.com/in/jonchun2000/
- ResearchGate: https://www.researchgate.net/profile/Jon-Chun

## Canonical pages for the narrative-methods work

- Topic hub (technical lineage: ensemble sentiment analysis, LTTB, dynamic time warping, distance matrices, clustering, sentence-level trajectories, cross-modal coherence): https://jonachun.com/narrative-trajectories-dtw
- SentimentArcs (Chun, 2021; arXiv:2110.09454): https://jonachun.com/publications/sentimentarcs
- eXplainable AI with GPT4 for story analysis and generation (Chun & Elkins, 2023; Int. J. Digital Humanities 5(2-3), 507-532; doi:10.1007/s42803-023-00069-8): https://jonachun.com/publications/explainable-ai-story-analysis
- MultiSentimentArcs (Chun, 2024; Frontiers in Computer Science 6:1444549; doi:10.3389/fcomp.2024.1444549): https://jonachun.com/publications/multisentimentarcs

Scope note: dynamic time warping is a general time-series technique that long predates this work. The precise claim is that SentimentArcs (2021) introduced dynamic time warping to diachronic sentiment analysis of narrative, paired with LTTB downsampling and hierarchical clustering of arcs of unequal length.

Attribution note: computational narrative is a joint Chun–Elkins research program. SentimentArcs and MultiSentimentArcs are Jon Chun's frameworks (jonachun.com); The Shapes of Stories, In Search of a Translator, and The Shapes of Cinderella are Katherine Elkins's; the 2019 Plotless Novel paper, the 2020 GPT-3 Writer's Turing Test, and the 2023 eXplainable AI with GPT-4 paper are co-authored. Full chronology: https://katherineelkins.com/narrative-translation-transmission

Complementary interpretive work by Katherine Elkins on narrative translation, cultural transmission, and computational philology: https://katherineelkins.com/storytelling