Student Preferences in Interacting with AI-Enhanced Learning Assistants (AIELA): A Comparative Study

AI-based virtual learning assistants are intelligent systems that revolutionize education by offering personalized learning opportunities that make learning methods more accessible, adaptive, and data-driven. These systems leverage machine learning and natural language processing to offer instant feedback and interactive engagement catering to individual student needs, thereby augmenting human learning assistants. This research compares two implementations of the AI-Enhanced Learning Assistant (AIELA) prototype, addressing limitations in scalability, privacy, and accessibility identified in the original Raspberry Pi-based implementation. The new VLA prototype implements a web application interface to replace the centralized hardware-based prototype in the previous model. The new model enables students to use their own mobile or computer devices to resolve specific hardware constraints. Two exploratory classroom demonstrations introduced the tool, followed by surveys gauging usability, engagement, and overall effectiveness. Students rated the web-based AIELA as user-friendly and moderately effective (~3.8-3.9), though its support for deeper learning and addressing misconceptions was lower (2.96). Despite comfort with using AIELA (4.15) and its usefulness for worksheets (3.73), HLAs consistently received higher marks for conceptual support, although limitations in survey design constrain direct comparisons. Future work should emphasize semester-long trials, improved data collection, and enhanced conversational strategies to balance Socratic prompting with direct guidance better, thereby complementing human instruction.