Tutor.AI

Basic data for this project

Description

The Tutor.AI project involves an interdisciplinary team consisting of professors, academic staff and the start-up colloc.AI developing an AI-based chatbot that is informed by academic content. The lecturers have adapted the knowledge base to their courses in order to provide the best possible and tailored support for the preparation and follow-up of the learning material. In education and academia, Large Language Models (LLM) such as GPT and services based on such technologies such as ChatGPT that use academic research in their context have gained much attention, although they also raise concerns such as cost, unknown bias and black-box behavior, as well as privacy concerns. In addition to ethical and didactic considerations, the use of generative AI in research and teaching also raises the question of dependence on commercial digital products - especially, but not exclusively, for AI applications. While universities have made significant developments in the field of AI during the decades of the so-called AI winter, they are now confronted with the accelerated development of AI technologies in the private sector. Tutor.AI uses advanced methods of natural language processing (NLP) and is operated on a university cloud infrastructure using Kubernetes. This enables a flexible and scalable application that runs in a privacy-compliant environment. The chatbot is able to efficiently process and generate context-related information, which increases the quality of interaction and the user experience. Tutor.AI will undergo a beta test in eight different courses this summer semester, which will be scientifically monitored and evaluated. Use is voluntary for students. The usage data will be stored anonymously exclusively for scientific purposes and for the further development of the tool.