Frequently asked questions
Factual answers about Jon Chun's research, methods, current projects, teaching, and engineering work.
Who is Jon Chun?
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 the Schmidt Sciences Humanities and AI Virtual Institute. Earlier he co-founded and led SafeWeb.
What was the Human-Centered AI curriculum at Kenyon?
Jon Chun and Katherine Elkins established the world's first Human-Centered AI curriculum and lab at Kenyon College in 2016. Jon designed the technical curriculum and courses; Jon and Katherine developed the broader interdisciplinary educational model together. The program gives humanities and social-science students enough technical fluency to conduct original computational and AI research while keeping disciplinary expertise at the center. Teaching and curriculum →
What is the 110–300× robustness paradox?
In studies of syntactic framing fragility, instruction-tuned large language models proved 110–300× more resistant to narrative manipulation than people, measured across healthcare, law, and finance vignettes. On this axis the models are far harder to fool than humans, a counterintuitive and measurable result about where LLM risk does and does not lie. See the research →
What is the confidence-scoring method for auditing language models?
Introduced in "Informed AI Regulation," it measures how firmly a model commits to a moral judgment versus hesitates, giving a way to compare normative certainty across models. It has since been applied across 1,613 social-decision scenarios (COLING 2025) and included among 69 foundational works in the AAAI 2026 "Beyond Verdicts" survey. It is co-authored with Katherine Elkins.
What is SentimentArcs?
Chun introduced SentimentArcs in 2021, a novel ensemble method for comparing narrative sentiment trajectories across texts. It evaluates dozens of sentiment models against one another and uses dynamic time warping to compare arcs of unequal length. The independently authored paper also reports that state-of-the-art transformers can struggle to find narrative arcs.
What is MultiSentimentArcs?
Chun introduced MultiSentimentArcs in 2024, the first multimodal method for measuring long-form film sentiment-arc coherence across modalities. It compares the trajectories recovered independently from a film's dialogue and images.
What did the eXplainable AI with GPT-4 paper contribute?
Chun and Elkins published the first application of explainable AI to narrative analysis in the International Journal of Digital Humanities in 2023. The paper develops sentence-level story trajectories and procedures for comparing narratives of unequal length on shared coordinates.
What is the through-line of the computational narrative research?
Building the instruments that make computational reading of narrative testable. "Middle reading" (2019) named a position between distant and close reading; SentimentArcs (2021) replaced the single trusted model with an ensemble; eXplainable AI with GPT-4 (2023) made story trajectories explainable and made narratives of unequal length comparable; MultiSentimentArcs (2024) carried the method across modalities, from text to film. Katherine Elkins's complementary program asks a different question: what stays the same and what changes when stories travel across languages, cultures, and time. See the novelty boundary →
What does dynamic time warping do for sentiment arcs?
Two novels rarely tell the same story at the same pace or the same length. Dynamic time warping (DTW) computes the distance between two whole arcs while absorbing the temporal shifts and stretches between them, so arcs that share a shape but not a tempo are recognized as similar. SentimentArcs (2021) introduced DTW to diachronic sentiment analysis, pairing it with LTTB downsampling (which reduces every arc to a common number of points while preserving its peaks, valleys, and endpoints) and using the resulting distance matrix to cluster arcs hierarchically. Across corpora running from roughly 1,400 to 13,000 data points per novel, this is what makes arcs of unequal length comparable by distance. See the novelty boundary → Narrative trajectories & DTW → Translation & cultural transmission (Elkins) →
What was 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 at the relevant historical point, built Triangle Boy, received the first security investment by In-Q-Tel, and was acquired by Symantec. Jon then served as director of development for Symantec's clientless VPN appliance line and holds two US patents related to browser-based SSL and clientless VPN appliances. SafeWeb and deployed systems →
What is Jon Chun's NIST CAISI role?
Jon Chun serves as Co-PI representing the Modern Language Association in the NIST CAISI consortium. His work includes LLM evaluation, red-teaming, and ethical auditing.
What is Archival Intelligence?
Archival Intelligence is part of the Schmidt Sciences Humanities and AI Virtual Institute. The project is building open computational infrastructure for rescuing endangered cultural archives in New Orleans, including Creole and Cajun multilingual newspapers and early jazz materials. As Co-PI, Jon contributes computational methods, AI workflows, and evaluation design for preservation, retrieval, provenance, interpretation, and access. How the system is being built →
Where can collaborators, journalists, and students start?
Journalists and the public: see Press and the featured findings. Academics and grant officers: see Research and Reception. Industry and investors: see Building. Students: see Teaching and the AI CoLab. Or get in touch directly.