Andrew Katz
· Associate ProfessorVerifiedVirginia Tech · Engineering Education
Active 1899–2026
About
Dr. Andrew Katz is an assistant professor in the Department of Engineering Education at Virginia Tech. He received his Ph.D. in engineering education from Purdue University in 2019, a master’s degree in environmental engineering from Texas A&M University in 2012, and a bachelor’s degree in chemical engineering from Tulane University in 2009. His research focuses on engineering ethics, decision-making, and system development, examining topics such as faculty mental models of engineering ethics and education, processes of change in ethics education, and students’ views of ethics and social responsibility. Dr. Katz is interested in understanding and improving the societal impacts of engineering and engineering education by studying how engineers, students, and faculty make normative decisions within their organizational and institutional contexts, aiming to inform better institutional designs within the engineering ecosystem. He has taught in both higher and basic education environments, including serving as faculty apprentice for an advanced research methods graduate course at Virginia Tech and developing and instructing a graduate course on Environmental Engineering Ethics and Society at Purdue University. Prior to his doctoral studies, he taught AP Physics at the Jesuit College Preparatory School in Dallas, Texas. Dr. Katz completed his doctoral studies through a Graduate Research Fellowship from the National Science Foundation and has received several awards, including the 2019 K. Patricia Cross Future Leaders Award, which recognizes promising future leaders in higher education committed to teaching and learning. His work aims to make the world a better place by helping others improve their work for societal benefit.
Research topics
- Computer Science
- Political Science
- Sociology
- Engineering
- Artificial Intelligence
- Engineering management
- Natural Language Processing
- Psychology
- Management
- Geography
- Civil engineering
- Medicine
- Mechanical engineering
- Library science
- Pedagogy
- Ecology
- Cartography
- Engineering ethics
- Medical education
- Social psychology
- Law
- Knowledge management
- Anthropology
- Mathematics education
Selected publications
Visualizing 30 Years of CHI Research with Generative AI
2026-04-13
articleOpen accessCHI has grown from roughly 60 papers annually in 1996 to over 1,000 in 2025, creating a corpus too large for manual thematic tracking. We present a methodology that combines Generative AI-assisted coding, text embedding, and multi-stage clustering to organize 11,847 CHI paper abstracts into 921 themes under 26 meta-themes. Bootstrap analysis confirms stability (Adjusted Rand Index [ARI] = 0.91) and 100% corpus coverage. Two complementary visualizations reveal distinct aspects of field evolution: a streamgraph showing absolute volume changes and a heatmap showing proportional shifts. Together, they distinguish growth from shifting priorities. Human-AI interaction increased by 60 times in absolute terms and by 5 times proportionally. At the same time, User-Centered Design Research grew modestly in absolute terms while its relative share fell to roughly a third, suggesting maturation into a foundational practice. These patterns, invisible at this scale, demonstrate how Generative AI-powered methods can reveal the dynamics of scholarly evolution.
Can LLMs Reason About Attention? Towards Zero-Shot Analysis of Multimodal Classroom Behavior
ArXiv.org · 2026-04-03
articleOpen accessUnderstanding student engagement usually requires time-consuming manual observation or invasive recording that raises privacy concerns. We present a privacy-preserving pipeline that analyzes classroom videos to extract insights about student attention, without storing any identifiable footage. Our system runs on a single GPU, using OpenPose for skeletal extraction and Gaze-LLE for visual attention estimation. Original video frames are deleted immediately after pose extraction, thus only geometric coordinates (stored as JSON) are retained, ensuring compliance with FERPA. The extracted pose and gaze data is processed by QwQ-32B-Reasoning, which performs zero-shot analysis of student behavior across lecture segments. Instructors access results through a web dashboard featuring attention heatmaps and behavioral summaries. Our preliminary findings suggest that LLMs may show promise for multimodal behavior understanding, although they still struggle with spatial reasoning about classroom layouts. We discuss these limitations and outline directions for improving LLM spatial comprehension in educational analytics contexts.
Humanities and Social Sciences Communications · 2026-01-21 · 3 citations
articleOpen access1st authorCorrespondingAbstract This work aims to answer one central question: to what extent can open-source generative text models be used in a workflow to approximate steps in thematic analysis in social science research? To answer this question, we present the Generative AI-enabled Theme Organization and Structuring (GATOS) workflow, which uses open-source machine learning techniques, natural language processing tools, and generative text models to facilitate aspects of thematic analysis. To establish evidence of validity of the method, we present three case studies applying the GATOS workflow, leveraging these models and techniques to inductively create codebooks similar to traditional procedures using thematic analysis. We show that the GATOS workflow can identify themes in the text that were used to generate the original synthetic datasets. We conclude with a discussion of relevant considerations, the implications of this work for social science research, and the tradeoffs of using open-source generative text models to facilitate scalable qualitative data analysis.
BMC Medical Education · 2026-01-31
articleOpen accessHealth systems science (HSS) education is an increasingly important component of undergraduate medical education. Despite curricular advances, the ways in which clinicians implement health systems science knowledge in everyday clinical practice, health systems thinking, remains understudied. A better understanding of how clinicians engage in health systems thinking to address everyday problems in clinical contexts is needed. We conducted semi-structured interviews with 10 expert clinicians experienced in undergraduate medical education, health systems science, and curriculum development to identify components of competent health systems thinking. Interview questions were informed by ecological systems theory and literature on learning professional competencies. Through interviews with experts, we have come to define health systems thinking (HST) as “an approach to solving problems in healthcare systems that utilizes a deeper understanding of interconnections and behavior of the entire system. As a skill, it coordinates the application of clinical and HSS knowledge and skills toward solving a contextual problem in the healthcare environment.” Clinician comments support the idea that HST is a metacognitive process rather than a specific subset of knowledge domains or affective attributes. This process requires that clinicians understand and navigate pressures on patient care originating from surrounding meso- and macro-systems. Medical students require more explicit exposure to HSS knowledge being implemented in clinical environments, and varied examples highlighting how meso- and macro-system patterns can impact individual patient care. This metacognitive integration of HSS knowledge into everyday clinical practice is critical for preparing medical students to meet the requirements of the accreditation council for graduate medical education (ACGME) core competencies in residency programs. Health systems thinking requires a method of operational assessment to provide students feedback and highlight targeted interventions for further development.
Extending Minimal Pairs with Ordinal Surprisal Curves and Entropy Across Applied Domains
arXiv (Cornell University) · 2026-03-15
articleOpen access1st authorCorrespondingThe minimal pairs paradigm of comparing model probabilities for contrasting completions has proven useful for evaluating linguistic knowledge in language models, yet its application has largely been confined to binary grammaticality judgments over syntactic phenomena. Additionally, standard prompting-based evaluation requires expensive text generation, may elicit post-hoc rationalizations rather than model judgments, and discards information about model uncertainty. We address both limitations by extending surprisal-based evaluation from binary grammaticality contrasts to ordinal-scaled classification and scoring tasks across multiple domains. Rather than asking models to generate answers, we measure the information-theoretic "surprise" (negative log probability) they assign to each position on rating scales (e.g., 1-5 or 1-9), yielding full surprisal curves that reveal both the model's preferred response and its uncertainty via entropy. We explore this framework across four domains: social-ecological-technological systems classification, causal statement identification (binary and scaled), figurative language detection, and deductive qualitative coding. Across these domains, surprisal curves produce interpretable classification signals with clear minima near expected ordinal scale positions, and entropy over the completion tended to distinguish genuinely ambiguous items from easier items.
Can LLMs Reason About Attention? Towards Zero-Shot Analysis of Multimodal Classroom Behavior
arXiv (Cornell University) · 2026-04-03
preprintOpen accessUnderstanding student engagement usually requires time-consuming manual observation or invasive recording that raises privacy concerns. We present a privacy-preserving pipeline that analyzes classroom videos to extract insights about student attention, without storing any identifiable footage. Our system runs on a single GPU, using OpenPose for skeletal extraction and Gaze-LLE for visual attention estimation. Original video frames are deleted immediately after pose extraction, thus only geometric coordinates (stored as JSON) are retained, ensuring compliance with FERPA. The extracted pose and gaze data is processed by QwQ-32B-Reasoning, which performs zero-shot analysis of student behavior across lecture segments. Instructors access results through a web dashboard featuring attention heatmaps and behavioral summaries. Our preliminary findings suggest that LLMs may show promise for multimodal behavior understanding, although they still struggle with spatial reasoning about classroom layouts. We discuss these limitations and outline directions for improving LLM spatial comprehension in educational analytics contexts.
SSRN Electronic Journal · 2025-01-01
preprintOpen accessSSRN Electronic Journal · 2025-01-01
preprintOpen accessSenior authorInternational Journal of Qualitative Methods · 2025-04-01 · 12 citations
articleOpen accessWe explore the possibility of using natural language processing (NLP) and generative artificial intelligence (GAI) to streamline the process of thematic analysis (TA) for qualitative research. We followed traditional TA phases to demonstrate areas of alignment and discordance between (a) steps one might take with NLP and GAI and (b) traditional thematic analysis. Using a case study, we illustrate the application of this workflow to a real-world dataset. We start with processes involved in data analysis and translate those into analogous steps in a workflow that uses NLP and GAI. We then discuss the potential benefits and limitations of these NLP and GAI techniques, highlighting points of convergence and divergence with thematic analysis. Then, we highlight the importance of the central role of researchers during the process of NLP and GAI-assisted thematic analysis. Finally, we conclude with a discussion of the implications of this approach for qualitative research and suggestions for future work. Researchers who are interested in AI-assisted methods can benefit from the roadmap we provide in this study to understand the current landscape of NLP and GAI models for qualitative research.
2024-08-03
articleOpen access1st authorCorrespondingThis paper describes the use of AI to support the initial development of an interview protocol designed to elicit engineering students' mental models of socio-ecological-technological systems (SETs) and how these models influence their design decisions.The protocol was created for a study that addresses the need to prepare engineering students to design sustainable solutions suitable for a world afflicted by climate change.Three frameworks informed the creation of the protocol: (1) mental models theory, (2) theory of planned behavior, and (3) social-ecologicaltechnological systems.Given advances in AI and the complexity of the theoretical frameworks, we were interested in learning whether generative AI could support protocol development.We generated questions using the generative text model: Claude-2.These generated questions were ranked by both Claude-2 and a member of the research team, and the rankings were compared.Through this process, we found that generative models can be used to write initial interview questions, but the quality of the questions is not consistent.Specifically, the questions generated were often relevant to the project, but they were not necessarily useful because of the use of awkward language.Despite this, the generated questions served as a helpful starting point for developing a large set of interview questions that were subsequently filtered and refined by the research team.
Recent grants
EAGER: Natural Language Processing for Teaching and Research in Engineering Education
NSF · $300k · 2022–2026
Frequent coauthors
- 47 shared
Isil Anakok
Virginia Tech
- 30 shared
Umair Shakir
Valley Tech Systems (United States)
- 30 shared
Justin L. Hess
Purdue University West Lafayette
- 28 shared
Holly Matusovich
Virginia Tech
- 27 shared
Homero Murzi
American Society For Engineering Education
- 22 shared
David B. Knight
Pennsylvania State University
- 20 shared
Brent Jesiek
Stanford University
- 19 shared
Kai Jun Chew
Kocaeli Üniversitesi
Labs
Education
PhD, Engineering Education
Purdue University
Awards & honors
- K. Patricia Cross Future Leaders Award (2019)
- Graduate Research Fellowship, National Science Foundation (2…
- PGSG Travel Grant Award, Purdue University (2018)
- Citizen Scholar Engagement Award, Virginia Tech (2016)
- NSF Water Scholarship, Texas A&M University (2010-12)
- Resume-aware match score
- Save to shortlist
- AI-drafted outreach
See your match with Andrew Katz
PhdFit ranks faculty by your research interests, methods, and publications — grounded in their actual work, not templates.
- Free to start
- No credit card
- 30-second signup