Learner Profiles
Responsible Use of AI Tools on HPC - Learner Profiles
Dr. Elena Vasquez - Humanities Faculty, Literature & Digital Culture
Role: Assistant Professor, English Department
Background: Elena is a second-year assistant professor specializing in 20th-century American literature and digital humanities. She has strong humanistic training and regularly works with texts, archives, and qualitative analysis methods. She’s completed a digital humanities certificate that touched on data analysis but never covered machine learning or AI. She has basic Python knowledge (self-taught from tutorials) and can read simple code but doesn’t maintain complex systems. She cares deeply about interpretive rigor and is attuned to how technology shapes meaning.
Current AI Experience: Elena has read widely about large language models and AI in the humanities: blogs, articles, conference panels. She’s played with ChatGPT and Claude for brainstorming and writing assistance but has never used them for research data. She understands that LLMs generate text by predicting likely continuations but is uncertain about training data, fine-tuning, and what happens to text she submits to cloud services. She knows generative AI exists but feels fuzzy on the technical details and worried about making mistakes.
Goals: Elena wants to use LLMs to analyze large corpora of literary texts and historical documents: identifying themes, tracing narrative patterns, and exploring intertextual connections at scales impossible through close reading. She’s excited by computational approaches that augment rather than replace interpretive scholarship. However, she’s deeply concerned about data privacy, intellectual property, and ethical questions around training data and environmental impact. She wants to understand Pomona’s AI policy, what data she can safely use with AI tools, and whether she should use local models (on HPC) or cloud services.
Special Considerations: Elena brings humanistic values of nuance, context, and ethical reasoning that are often missing from technical AI discussions. She’ll be initially skeptical of claims that LLMs “understand” text and wants to understand limitations. She learns best through concrete literary examples and clear explanations of how LLMs work conceptually (not heavy math). She’ll be highly motivated by understanding responsible, ethical use and how to design research that interrogates rather than uncritically accepts AI outputs. Connecting AI tools to her existing interpretive frameworks and emphasizing that careful, critical thinking is a humanistic strength she can bring to AI research will build her confidence.
Dr. Marcus Liu - Chemistry Researcher, Drug Discovery Focus
Role: Senior Research Scientist, Chemistry Department
Background: Marcus has a PhD in computational chemistry and 6 years of experience using computational methods for molecular modeling and property prediction. He’s deeply familiar with scientific computing, has programmed in Python and C++, and regularly works with proprietary molecular datasets from industrial collaborations. He understands machine learning at a conceptual level and has used scikit-learn for property prediction. He’s security-conscious about proprietary data and regularly handles confidential collaborator information under NDAs.
Current AI Experience: Marcus has heard that generative models and large language models are being applied to drug discovery: for molecular generation, protein folding prediction, and retrosynthesis planning. He’s read papers on AI-driven drug discovery and is curious about fine-tuning models on his proprietary datasets to predict molecular properties specific to his company’s chemistry. However, he’s genuinely worried about data security: his datasets contain proprietary chemical structures, synthesis details, and test results that could be worth millions if leaked. He’s never uploaded data to cloud AI services and is skeptical about their security guarantees.
Goals: Marcus wants to leverage recent AI advances (transformer models, molecular foundation models) to accelerate drug discovery in his research group. Specifically, he wants to fine-tune pre-trained models on his proprietary datasets to improve molecular property predictions, explore generative models for novel compound design, and potentially use LLMs for literature mining and assay design. Critically, he needs to do this safely: understanding how to work with sensitive data locally on HPC systems rather than uploading to cloud services, ensuring compliance with industry collaborator agreements, and maintaining intellectual property protection.
Special Considerations: Marcus is technically sophisticated, security-minded, and risk-aware. He understands proprietary data sensitivity and will push back on any cloud-based solution that isn’t ironclad secure. He’ll respect local HPC solutions that avoid external data transfer. He doesn’t need basic ML education but wants guidance on Pomona’s data classification system, local vs. cloud trade-offs for sensitive data, and how to set up safe workflows on HPC. He’ll be highly motivated by practical solutions that solve his security concerns. Emphasizing local model hosting, data classification frameworks, and compliance mechanisms will build his trust and engagement.
Alexis Johnson - Economics Student, Data-Focused
Role: Junior Economics Major, Research Interest in Labor Markets
Background: Alexis is a third-year economics major with a strong quantitative focus. She’s completed econometrics courses using R and Stata, understands regression analysis well, and is comfortable with empirical research methods. She recently discovered that economists are using machine learning and neural networks for causal inference and policy analysis. She’s self-taught in Python (one semester course plus online tutorials) and is eager to learn modern computational methods. However, she’s primarily worked with published, anonymized datasets and isn’t familiar with human subjects research protocols or data privacy regulations.
Current AI Experience: Alexis has never used an LLM or fine-tuned a machine learning model. She’s seen popular articles about AI in economics and wants to understand how to use these tools in her research. She has questions about what LLMs can do for economic analysis, whether they can help analyze survey data, and whether she can train models on her own research datasets. She doesn’t fully understand FERPA (Family Educational Rights and Privacy Act) requirements for student data or the restrictions on handling survey responses from human subjects. She’s enthusiastic but potentially unaware of important legal and ethical constraints.
Goals: Alexis wants to fine-tune language models on survey data she’s collecting about labor market perceptions and career choices among college students. She’s interested in sentiment analysis, automated coding of open-ended responses, and understanding what her survey respondents really think about policy changes. She also wants to use transformer models for text classification of job postings to track labor market dynamics. However, she needs to learn about data privacy regulations that apply to her work: specifically FERPA constraints around student data, Institutional Review Board (IRB) requirements for human subjects research, and how to handle survey responses responsibly.
Special Considerations: Alexis is bright, motivated, and technically self-directed but lacks experience with regulated data and human subjects research constraints. She’ll benefit from education about why data privacy matters and how regulations protect vulnerable populations. She may initially resist or not understand privacy restrictions (“but I have permission from respondents”), so education needs to emphasize both legal requirements and ethical responsibilities. She learns well by doing but needs clear guardrails and oversight. Connecting privacy concerns to economics (e.g., how data breaches harm individuals, historical data misuse in social science), explaining FERPA and IRB requirements clearly, and showing her how to work responsibly with sensitive data will help her develop as an ethical researcher while still pursuing her interests.
David Park - IT Staff, Policy & Governance
Role: Information Technology Specialist, System Administration & Compliance
Background: David has 5 years of IT experience at Pomona, primarily managing systems, security, and infrastructure compliance. He has solid technical knowledge of Linux systems, networking, and data center operations. He understands institutional policy frameworks and has implemented various security and compliance measures. He’s becoming increasingly involved in AI governance as the institution develops policies around AI tool usage and data security. He has general knowledge of AI systems but limited hands-on experience and wants to deepen his understanding to enforce and advise on institutional policies effectively.
Current AI Experience: David understands AI conceptually from reading institutional policy documents and security advisories, but has never trained a model or worked with LLMs in depth. He knows that AI tools exist and that they raise security concerns, but he’s uncertain about specific technical risks: what data is retained by cloud AI services, how to classify data for safe AI usage, what HPC-based AI tools can do, and how to detect misuse. He’s tasked with helping develop Pomona’s AI governance framework and needs to understand the landscape well enough to write informed policies.
Goals: David wants to develop and enforce Pomona College’s responsible AI policies across the institution. Specifically, he needs to understand the AI landscape (what AI tools exist, how they work at a high level), what data security and privacy risks different approaches present, how to classify institutional data appropriately for AI usage, what technical controls can help enforce responsible AI practices, and how to educate faculty and staff about compliant usage. He also needs to work with the HPC team to understand how to safely host AI tools locally and what monitoring/oversight mechanisms are appropriate.
Special Considerations: David brings IT governance expertise and cares about institutional risk management. He doesn’t need ML theory but needs practical understanding of security implications and technical capabilities. He’ll value clear frameworks (data classification schemes, risk matrices), concrete technical controls (monitoring, access management), and policy templates he can adapt. He works best with specifics over abstractions. He’ll be highly motivated by practical tools to enforce policy and by solving real governance challenges. Providing data classification frameworks, technical controls and monitoring approaches, risk assessment models, and concrete policy templates will help him succeed. He’s also a critical stakeholder: his understanding and support of responsible AI policies will influence institutional implementation significantly.