AI-supported learning at the Ohm
In the joint project HAnS, AI is being made usable for students' everyday learning.
Since 2021, an interdisciplinary research network led by Nuremberg Institute of Technology Georg Simon Ohm (Ohm) has been working on making artificial intelligence (AI) usable for students' everyday learning. Just over a year after the introduction of ChatGPT, the world of education is upside down. There is hardly a question that the Large Language Model (LLM) cannot answer. Data protection and copyright are more than questionable in this context. As part of the joint project HAnS (the intelligent university assistance system), Ohm is developing a learning experience platform that makes LLMs usable in compliance with data protection and copyright law for the first time. This development is being accompanied didactically and evaluated scientifically.
Surveys have shown that many students like to repeat content from lectures in order to consolidate it. This is countered by a new form of learning in which you engage in a dialog with a chatbot, such as ChatGPT. The special feature of the HAnS approach is to create a structured and fact-based teaching/learning environment by linking the LLM world knowledge with didactic materials. Lecturers can also specifically incorporate this possibility into their didactic concept.
Prof. Dr. Robert Lehmann (Faculty of Social Sciences) is one of the lecturers involved in the test run. "I work a lot with the flipped classroom method, which requires students to work on materials independently before the respective teaching unit," he explains. He had already created teaching videos for this purpose before HAnS. "HAnS now enables students to find their way around the teaching videos even faster, read up on individual passages in the transcripts and enter into dialog with the teaching material using a local open-source LLM-based chatbot. This can intensify engagement with the content." The research focuses on the question of how to differentiate between general world knowledge and the available material.
Lehmann believes that didactic support for the system is crucial for learning success. This is because there is also a risk that students will primarily access the AI-generated summaries. "Here it is necessary to use suitable didactic interventions to support the motivation for extensive study." Planned functions such as the automatic creation of multiple-choice questions or more extensive exercises on the content offer interesting possibilities here. The development of the platform is specifically geared towards the needs and expectations of students and lecturers, which are scientifically determined during the process.
Ohm is responsible for the overall management, technical development and accompanying evaluation in the HAnS joint project. Other partners in the project are responsible, for example, for clarifying data protection, ensuring optimum acceptance of the system or ethical and copyright issues. Different specialist areas work together on an interdisciplinary basis: Findings from the fields of software architecture, machine learning, language and text comprehension, didactics, social science, educational science, computer science, law and ethics are brought into relation with each other and incorporated into the gradual development and revision of HAnS. The aim of the project by 2025 is to develop a didactically and scientifically sound open source solution that is also available for use at other universities.
The HAnS project is funded by the Federal Ministry of Education and Research (BMBF) until 2025 as part of the federal-state initiative "AI in Higher Education" and the funding priority "Digital Higher Education".
As part of the HAnS project, a research team at the Ohm is developing a data protection and copyright-compliant learning experience platform. Photo: TH Nuremberg | Screenshot from research film
