Navigation auf uzh.ch
Navigation auf uzh.ch
Aidviser is an educational tutoring chatbot developed by the Teaching Center at the Department of Finance, University of Zurich. Initiated in Spring 2024 through a strategic collaboration with the Executive Education Finance of the Faculty of Business, Economics, and Informatics (WWF), Aidviser represents one of the first course-specific AI tutoring implementations at UZH, predating many centralized AI initiatives. The project addresses the pedagogical challenge of providing personalized learning support in large-enrollment courses.
The primary objective of Aidviser is to offer students continuous access to course-specific guidance and explanations, effectively functioning as a supplementary tutor available outside regular teaching hours. By leveraging artificial intelligence technologies, the chatbot creates a scalable approach to providing individualized support that would otherwise require resources beyond what is typically available in university settings with hundreds of students.
The Aidviser chatbot is built on sound pedagogical principles that address the challenge of providing personalized learning support at scale. The foundational concept recognizes that effective learning often requires individualized guidance, particularly in complex subjects like finance. By functioning as a supplementary 1:1 tutor, Aidviser creates a supervision ratio that would be impossible to achieve through traditional means in courses with hundreds of students.
The pedagogical approach emphasizes just-in-time learning support, allowing students to receive assistance precisely when they encounter difficulties in understanding course concepts. This immediate feedback loop is crucial for maintaining student engagement and preventing the accumulation of knowledge gaps that can impede further learning. For first-semester students in their assessment year, this continuous support is particularly valuable as they navigate the transition to university-level finance concepts.
Aidviser implements an AI-enhanced tutoring approach that combines the accessibility of digital tools with personalized learning interactions. Unlike generic AI assistants, Aidviser is focused on specific course materials, enabling it to provide contextually relevant support aligned with the instructor's teaching approach and terminology. The system is designed to supplement rather than replace human instruction, offering additional practice opportunities and concept clarification outside of regular teaching hours.
The 24/7 availability of the chatbot addresses a critical need for flexible learning support, particularly during exam preparation periods when students most actively engage with course materials. This constant availability ensures that learning momentum isn't disrupted when students encounter conceptual obstacles outside of scheduled teaching hours.
The learning design of Aidviser is guided by several key principles:
Aidviser employs a flexible multimodel architecture that supports integration with multiple AI providers, including OpenAI, Anthropic, Gemini, TogetherAI, and others. This design choice enables students to select the most appropriate model for their specific needs. Different AI models offer varying capabilities in terms of reasoning depth, contextual understanding, and specialized knowledge. The system connects to chosen providers through standardized APIs, maintaining a flexible infrastructure that can adapt as AI technology evolves.
The chatbot offers distinct interaction modalities to support different learning needs. In tutoring mode, rather than providing direct answers, Aidviser guides students through problem-solving processes, encouraging critical thinking and deeper engagement with course material. When students need immediate clarification, the explainer mode offers clear, direct explanations with an emphasis on conceptual understanding. Students can actively switch between these modes depending on their learning needs, fostering autonomy and personalized learning experiences.
To ensure responsible resource use, Aidviser implements a credit management system that allocates specific usage limits per student. The credit system serves an educational purpose by making students aware of the varying costs associated with different AI models and interaction patterns, helping prepare them for responsible use of AI resources in professional contexts. This approach ensures equitable distribution of AI resources across all enrolled students. Usage data collected through this system provides valuable insights into student engagement patterns, informing both technical refinements and pedagogical adjustments.
The Teaching Center conducted comprehensive evaluations of Aidviser through surveys in both Spring and Fall 2024 semesters, including dedicated assessments and questions within the standard course evaluations (LVE). Additionally, a focus group with course tutors was organized in Fall 2024 to gather insights from their perspective. In Spring 2025, we are conducting student focus groups to explore AI chatbot usage patterns and preferences in greater depth.
Student feedback has been predominantly positive among those who engaged with the chatbot. Particularly appreciated features included the chatbot's focus on course-specific content without introducing external material, the tutoring mode's approach of guiding rather than immediately providing solutions, and the inclusion of lecture transcripts that covered administrative matters and current developments. Students also valued the ability to switch between different language models, noting the educational benefit of experiencing varying capabilities across different AI providers—a skill increasingly relevant in professional contexts.
Some constructive feedback highlighted areas for improvement, such as occasional hallucinated references, the inability to upload files for specific exercise assistance, and the sometimes rigid conversation flow in tutoring mode. Interestingly, a subset of students reported not using the chatbot due to general skepticism about AI accuracy, preferring to rely directly on official course materials—a perspective that informs our ongoing development of trust-building features and transparent source attribution.
The development of Aidviser has followed a methodical, evidence-based approach across several key phases:
Spring 2024: Initial pilot implementation in Banking and Finance II, where the chatbot was introduced as an experimental tool later in the semester. Despite limited availability based on a credit system, students engaged extensively with the platform, generating over 3,400 interactions in just a few weeks. This pilot provided crucial insights into student usage patterns and technical requirements.
Fall 2024: Large-scale deployment to over 900 students in Banking and Finance I, supported by UZH Teaching Fund (ULF) micro_innovation funding. This implementation operated under a "fair use" model throughout the entire semester, allowing students what felt like unlimited access while maintaining responsible resource management behind the scenes. Student feedback was systematically collected through surveys and workshops to inform future improvements.
Spring 2025: Current phase focusing on expanding Aidviser's capabilities to support more complex disciplinary contexts. We are developing and testing advanced use cases for coding and mathematics courses, while implementing an agentic platform architecture to enhance the system's flexibility for the planned Summer 2025 integration into KlickerUZH. Additionally, we are collaborating with central UZH projects like AI Buddy to investigate future potential and ensure alignment with university-wide AI initiatives.
Summer 2025: Planned integration with KlickerUZH, the classroom interaction platform developed at UZH. This integration will create a comprehensive digital learning ecosystem that combines real-time classroom interaction with persistent AI-enhanced tutoring support. For detailed information about this integration and the technical implementation, please refer to the KlickerUZH use case on chatbot tutoring.
Each phase has contributed valuable insights to the project's evolution, with comprehensive findings documented in the KlickerUZH use case. This iterative process has allowed for continuous improvement while maintaining focus on the core pedagogical objectives that motivated the project's inception.
In Spring 2025, we are actively exploring and testing more advanced applications of Aidviser, particularly in coding and mathematics courses that require specialized problem-solving approaches. The implementation of an agentic platform architecture is enhancing the system's flexibility and capability to handle these more complex disciplinary contexts, establishing the technical foundation needed for the planned Summer 2025 KlickerUZH integration.
Our research and development roadmap focuses on several key areas: enhancing the chatbot's ability to provide meaningful feedback on student work, developing more sophisticated tutoring strategies for different disciplines, improving the accuracy of source attribution, and optimizing the balance between accessibility and resource management. We are also investigating how usage data can inform teaching practices and curriculum development.
The integration with KlickerUZH will facilitate scaling Aidviser across departments and disciplines at UZH. We are developing templates and guidelines to help instructors in different fields adapt the chatbot to their specific needs, ensuring that the technology remains pedagogically sound as it expands beyond its original finance context.
The evolution of Aidviser from a standalone chatbot to an integrated component of KlickerUZH represents a natural progression in our educational technology development. As the creators of both Aidviser and KlickerUZH, the Teaching Center at the Department of Finance is uniquely positioned to ensure seamless integration between these complementary tools. This integration will make AI-powered tutoring capabilities available to all lecturers across UZH for their students, significantly expanding the impact of our initial innovation.
The original collaboration with the Executive Education Finance, WWF provided the foundation that made this integration possible. By allowing us to develop and test Aidviser as a standalone tool first, we gained crucial insights into student usage patterns, technical requirements, and pedagogical approaches that now inform the KlickerUZH integration. This phased development approach has enabled us to refine the core functionality before scaling it to a broader platform implementation.
The integration process presents both technical and pedagogical challenges. On the technical side, we are working to ensure that the chatbot functionality operates seamlessly within the KlickerUZH environment while maintaining the flexibility of our multi-model architecture. Pedagogically, we are developing guidelines and templates to help instructors across disciplines effectively implement AI tutoring in their specific contexts, recognizing that different subjects may require tailored approaches to AI-enhanced learning support.
The Aidviser project represents a significant step forward in our approach to supporting student learning at scale. From its inception through collaboration with Executive Education WWF to its upcoming integration with KlickerUZH, this educational chatbot has demonstrated the potential of AI to enhance the learning experience while maintaining sound pedagogical principles.
The journey from a standalone tool to an integrated platform component illustrates our commitment to iterative improvement based on evidence and student feedback. With support from the UZH Teaching Fund (ULF) micro_innovation grant, we've been able to provide equitable access to AI-powered tutoring for hundreds of students, gathering valuable insights that will inform future implementations across the university.
As we continue to develop more advanced use cases and prepare for integration with KlickerUZH in Summer 2025, we remain focused on our core mission: creating innovative, accessible, and pedagogically sound learning experiences that prepare students for a future where AI literacy will be increasingly important. Through ongoing collaboration with central UZH services and knowledge sharing across the institution, the Teaching Center at the Department of Finance continues to contribute to UZH's position at the forefront of educational innovation.
For more information about Aidviser or to discuss potential collaborations, please contact the Teaching Center at the Department of Finance.
