Projects
The Rectorate is undertaking a number of projects in the field of AI in education and is promoting initiatives on AI-based innovations for teaching in the ETH departments.
Innovedum projects on AI
As part of the Innovedum Fund for the promotion of innovative teaching at ETH, projects in the Innovedum focus area of AI in teaching and learning are currently being supported. Projects are being funded that deal with the integration and application of existing AI tools in teaching.
- Assessing the Potential of AI for Scientific Writing Techniques: The focus is on the use of GPT bots and applications with large language models for teaching academic writing. Teaching modules on AI in academic writing, AI and plagiarism, AI and research integrity are being developed.
- Application of Large Language Models (LLM) in Digital Engineering: The focus is on developing skills for the reflective use of AI-based tools for time-efficient programming and their use as a source of inspiration.
- Evaluating the impact of AI-based tutoring systems on student learning in biology education: This project investigates the effectiveness of an AI chatbot on the learning behaviour and success of students in biology education.
- PIA - Privacy Implications of AI-based Personalized Learning Tools: The project provides a personalised AI-based tool that makes privacy implications tangible and allows students to not only reflect on but also control the degree of privacy intrusion.
- AI-aided programming and collaborative software development: AI-assisted scientific programming in Python using Github for project deployment using version control and synchronous collaboration.
- Explore the responsible use of AI in generating scientific texts, images, audio and code: How can early career scientists gain experience to use generative AI responsibly in science.
AI-based projects for teaching
The following are some projects that the rectorate is currently conducting. If you are your department is interested in working with us, please contact us. Please also note that such projects can be supported through the Innovedum focal point theme AI in Teaching and Learning.
- Problem Solving: exploring the problem-solving capabilities of AI in STEM courses. For example, external pageGPT can pass introductory physics coursescall_made.
- Accessibility: using artificial intelligence be used to make lecture materials accessible for learners with disabilities. For example, external pageLaTeX documents and slidescall_made. Another topic is generating ALT-tags for images, and using AI to generate transcripts of blackboards.
- Grading: using artificial intelligence to assist in grading handwritten mathematical derivations and proofs. So far, external pagean agreement of R2>0.8 could be achieved between manual and AI gradingcall_made. Also, a pre-trained GPT-4 has been shown to perform as well as specificially trained hand-engineered system for external pageautomatically grading short answers to science questionscall_made.
- Course Evaluations: artificial intelligence may help in more quickly gaining an overview of open-ended course evaluations.
- Usage by Learners: survey on awareness regarding opportunities and challenges of AI-tools, as well as usage patterns among learners at ETH Zurich.
- Retrieval Augmented Generation: increasing the accuracy of AI-generated chats by retrieving relevant sections of course materials and inserting them into the prompts.
- Spezialized Large Language Model: in collaboration with EPFL, we plan to tune an open-weight Large Language Model for our teaching mission. We appreciate contributions of teaching materials for tuning purposes.
Several of these projects are pre-studies for Project Ethel.