Conference Program

AI Assistant Show and Tell - Gasson 112

• Tim Lindgren, Tim Lindgren, Assistant Director for Design Innovation, Center for Digital Innovation in Learning 
• Ashley Duggan, Professor, Health Communication
• Marta Mirete Hernandez, Lecturer of Spanish; Coordinator of Intermediate Spanish Language Program
• Seoyeon Bae, Ph.D. Candidate
• Chris Wilson, Professor Emeritus, English
• Stefane Cahill Farella, Senior Associate Director, Employee Development, Human Resources
• Callid Keefe-Perry, Assistant Professor, Clough School of Theology and Ministry
• Kyle Fidalgo, Academic Technologist, Law School

Join us for an interactive “Show & Tell” session featuring a range of custom AI Assistants developed by members of the Boston College community. Attendees are invited to drop by, ask questions, try out the chatbots, and learn firsthand from the creators about their design choices, lessons learned, and future ideas.

POSTER SESSION  - Gasson 100 & Gasson 112

Utilizing LLM tools to engage non-STEM majors in learning to use prompting to code

• Sheikh Ahmad Shahm, PhD student

This poster is about how Large Language Model (LLM) chatbots like Copilot can support non-STEM undergraduates with limited coding experience. Three lab sessions were conducted focusing on creating JavaScript-based games and simulations, where students used LLM tools to create progressively more complex programs, the final one being their own project. While survey results were mostly inclusive, students’ chat history and reflections suggest growth in students’ ability to write better prompts to generate better code. However, students had less critical engagement with code itself. The study provides early insights into the role of AI tools in coding education among novice learners.

Artificial Intelligence Through the Lens of Critical Information Literacy

• Library Instruction Department

This poster conveys the importance of critical information literacy and the library's role in supporting students' learning and research processes.

Enhancing Neonatal Muscle Segmentation with Mixed Phantom and Clinical Ultrasound Data

• Nora Alwash, Class of 2026, Undergraduate Research Assistant
• Hayoung Cho, Class of 2025, Research Student
• Vera Hernandez, Class of 2026, Undergraduate Research Assistant
• Mary Loeb, Class of 2026, Undergraduate Research Assistant
• Bryan Ranger, Ferrante Family Assistant Professor, Department of Engineering

Monitoring neonatal body composition non-invasively can detect malnutrition and growth patterns, critical for preterm infants. We used abdominal, biceps, and quadriceps ultrasound images to measure tissue thickness, predicting fat-free mass (FFM) and fat mass (FM) via non-linear regression methods and Gaussian Process Regression. Sensitivity analysis showed individual measures were insufficient, prompting UNet-based cross-sectional area segmentation. We developed gelatin-based phantoms (1% psyllium husk for adipose; 2% aluminum oxide in agar for muscle) with echogenicity comparable to clinical datasets. Dice scores were 0.9272 for phantom, 0.6089 for clinical, and 0.8027 for mixed data. Ongoing work focuses on improving predictive performance on mixed and clinical datasets. 

“AI gets a reputation for making interactions feel surface-level, but it doesn’t have to be that way”: Students’ reflections on using purposeful AI to facilitate reflective conversations

• Maddie Bruns, Class of 2025
• Annette Choi
• Aimee Cowles, Class of 2026
• Hannah Cunniffe, Class of 2026
  
• Ashley Duggan, Professor, Health Communication
• Lily Harden, Class of 2025
• Aliza Jernigan, Class of 2025
• Erin Kolenda, Class of 2026
• Charlotte LaBossiere, Class of 2025
• Frances Lee, Class of 2026
• Croía Loughnane

As digital technologies increasingly shape how we communicate, learn, and care for ourselves, there is growing potential to use artificial intelligence (AI) to support reflection, connection, and wellbeing. AI chatbots, in particular, are emerging as tools not just for delivering content, but for facilitating transformative conversations—the kinds that spark awareness, build empathy, and inspire action. Two custom chatbots were designed to be thoughtful, curious, and facilitative conversations, drawing inspiration from person-centred health and wellbeing professionals with strong communication skills. This study employed a qualitative exploratory design to investigate the effectiveness of the reflective AI chatbot within the context of a university course on Health and Illness in Relationships at Boston College. This poster focuses on themes identified in student written reflections about their AI chatbot experience.

Harvesting Deep Networks: Towards Designing a Robust Crop-yield Estimation Model 

• Shrey Gupta, Post Doctoral Research Fellow, Computer Science
• Yi Ming, Institute Professor of Climate Science and Society, Schiller Institute for Integrated Science and Society
• George Mohler, Daniel J. Fitzgerald Professor, Computer Science

Accurate crop yield prediction is vital for food security and agricultural planning, yet building robust models that handle the complexities of temporal and spatial variations remains a challenge. This study introduces a deep learning framework employing a hybrid Convolutional Neural Network–Long Short-Term Memory (CNN-LSTM) architecture to model both temporal dynamics and spatial dependencies in long-term agricultural data. Trained on multi-year, county-level crop yield data alongside daily meteorological variables (temperature, precipitation, humidity, and more), the model captures intricate patterns across time and space.

The LSTM component effectively learns long-range temporal dependencies and seasonal fluctuations, which are crucial for accurate forecasting. Simultaneously, the CNN component extracts localized spatial features from county-level inputs, enabling the model to understand spatial patterns in climate and their impact on yield variations across regions.The results demonstrate that this CNN-LSTM framework achieves accurate yield predictions while effectively capturing the complex interactions between long-term climatic trends and spatial differences in county-level crop yield data. This hybrid approach offers a significant step towards more reliable and insightful crop yield forecasting.

Divine Guidance or Digital Counsel? Investigating the Relationship Between Religiosity and AI Advice-Seeking

• Helen Huiting Zheng, PhD Candidate, Psychology
• Kyle Fiore Law, Postdoctoral Researcher
• Liane Young, Professor, Psychology 

In recent years, it has become increasingly common for people to seek advice, including moral guidance, from artificial intelligence (AI) chatbots. This poster presents findings from a representative U.S. adult sample (N = 348) examining the relationship between individuals’ religiosity and their tendency to seek advice from AI chatbots for both general and moral matters. The findings offer initial insights into how religious involvement and commitment shape the tendency to seek advice from AI chatbots. This work contributes to understanding the intersection of technology, belief systems, and advice-seeking behavior in contemporary society.