Background Information
Generative AI tools such as ChatGPT have become widely used in higher education, offering students quick access to information and assistance. However, most existing tools draw on generic web content and large-scale pretraining, which can result in inaccurate or contextually irrelevant answers, particularly problematic in academic settings. Educators need AI support tools that are aligned with specific course content and structured learning goals. The SAIL project addresses this need by piloting the chatbot Alexandria, a Retrieval Augmented Generation (RAG)-based tool that sources only from approved course materials. This ensures reliable, focused support for students and keeps teachers in control of content and assessment.
The project will run across eight courses at TU/e, covering disciplines from behavioral science and chemistry to data science and automotive engineering. Each course will integrate the chatbot, aiming to improve student autonomy, motivation, and learning outcomes. SAIL also emphasizes responsible AI use, collaborating with LIS to ensure full compliance with data privacy and security standards. A robust mixed-methods evaluation will gather feedback from both students and instructors to assess effectiveness, usability, and educational impact. Ultimately, SAIL contributes to a more inclusive, flexible, and future-proof learning environment that aligns with TU/e’s strategic goals.
Aim of the project
The main goal of SAIL is to explore how AI chatbots can be securely and effectively integrated into TU/e’s education to enhance student learning while safeguarding academic integrity and privacy. The project aims to move beyond unsupervised use of generic AI tools by developing a structured, course-specific implementation of AI that supports self-directed learning and aligns closely with each course’s objectives and materials.
By piloting the Alexandria chatbot across eight diverse courses, SAIL seeks to understand how different types of students engage with AI, how it affects their motivation, learning outcomes, and group collaboration, and how it may reduce instructors’ workload. The chatbot is designed not just to answer questions, but to encourage deeper engagement through Socratic dialogue and guided feedback, acting as a digital teaching assistant that’s available anytime.
Simultaneously, SAIL investigates the technical and organizational infrastructure needed to support such a tool responsibly. This includes addressing privacy concerns, designing integration with the university’s LMS, and exploring future options for a fully compliant, locally hosted TU/e-specific AI solution. The insights from this pilot will inform long-term policy and development strategies for AI-enhanced learning across the university.