Student chatbot use and perceptions in a course assignment: Comparing two geomatics courses

Recent advancements in generative AI have significantly impacted higher education, especially through the integration of AI chatbots in course design, teaching, administration, and student support. However, empirical research on chatbot use in Geomatics education remains limited, particularly regarding students’ choice of platforms, preferences for multimodal prompts, perceived quality of chatbot responses, challenges encountered, and follow-up strategies. To address this gap, we investigated how students used chatbots as part of a given discussion assignment in two university-level Geomatics courses offered in Spring 2025: Measurement Science (n = 20) and Geospatial Analysis (n = 25). As part of the discussion assignment students were tasked to design five prompts based on lecture content, submit them to one of seven chatbot platforms, rate the responses, and provide written justifications. Student interactions were analyzed using regression models that accounted for repeated prompts per student. These models examined how platform choice, prompt modality, and other prompt characteristics influence response ratings and follow-up behavior. Findings revealed distinct patterns. For example, students in Geospatial Analysis posed proportionally more conceptual than computational questions; follow-up prompts were more likely when chatbots produced computational errors than when explanations were unsatisfactory; and data-enhanced prompts (e.g., with tabular inputs) received lower ratings than textual ones, underscoring unmet expectations for chatbot-based GIS data processing. The study concludes with a set of best practice recommendations for chatbot use based on these two courses, which may inform Geomatics instruction in similar university-level courses. • Students rated chatbot responses more favorably for surveying-related prompts than for GIS-related prompts • Multimodal prompts encouraged greater curiosity and exploration among students • Computation-focused prompts produced more unsatisfactory responses and follow-up questions than conceptual prompts • Students used a range of follow-up strategies when initial responses were unsatisfactory • Guidance is needed to help students develop advanced prompting techniques