More Effective Teaching, Easier Learning - AI Projects At BME

"The goal is clear: to relieve the burden on lecturers, so that they can spend less time on administrative or repetitive tasks and more time on what really matters - students, research and quality professional work," is how Hassan Charaf summed up what he expects from projects to promote and regulate the use of AI at the university, in an interview published in the December issue of “Műhely”.

And there are plenty of such projects at BME: from third-party products to in-house development to the purchase of necessary hardware, there are at least half a dozen programs running that can hopefully make the work of students and lecturers easier and more efficient. We had a look at the tools with Tamás Mészáros, a member of the Artificial Intelligence Research Group at BME.

A trial of GoSchool, an AI-based tool provided by a third-party provider, has been running since the beginning of September, with around 20 lecturers and 100 students participating, but anyone interested can still join. The system, called edAI in the BME environment, is essentially a web chatbot that is trained by the lecturers by uploading the course material. Students can ask questions and the chatbot will answer them in the style of the lecturer. The questions can later be retrieved anonymously, so the lecturer can see where further explanation or clarification is needed. And learning becomes a continuous dialogue instead of a solitary struggle.

BME also encourages innovative solutions from its own lecturers, so in autumn the university launched a call for developers to design and develop AI-based tools, systems and prototypes to improve the efficiency and quality of education. The following three applications have been awarded funding.

Bertalan Forstner (VIK, Department of Automation and Applied Informatics) is developing a system for generating exam questions, evaluating answers and providing detailed feedback using large language models. The aim is to provide students with immediate and detailed feedback and to save lecturers time in preparing and improving exams.

Evelin Bell (Faculty of Chemical Technology and Biotechnology, Department of Organic Chemistry and Technology) uses existing notes to create structured course summaries, which the AI uses to generate questions for students to check their own knowledge.

The main objective of the development is reducing the drop-out rate.

Gábor István Hullám (VIK, Department of Artificial Intelligence and Systems Engineering) is working on the implementation of a gamification incentive system that gives feedback to the student, showing which areas need extra attention. The role of AI is to analyse performance and recommend new assignments based on this, but it can also give an indication if the student is behind on a pro rata basis - this can be particularly useful for compulsory assignments. The three projects are expected to be completed by spring.

Analitycal solutions

On behalf of the Directorate of Education, the VIK experts are working on an AI-based analytical tool from which lecturers could draw analyses based on subject data to identify problem areas. Results in this case are also expected next spring, but in the meantime the Directorate of Education is also looking at third-party solutions with similar functionality, such as IntelliBoard.

Several Microsoft AI tools are already in use at BME, but on a relatively small scale. Members of the Artificial Intelligence Research Group have recently agreed with the company's European and national leaders to strengthen cooperation and awareness of the tools. Microsoft experts will soon be presenting best practices and ready-to-use educational solutions to BME's lecturers at an information day.

In addition to educational developments, there is also a focus on AI in many training programmes and laboratories, so the hardware background is important. Procurement of accelerator cards and target hardware for running and customising AI models is under preparation and they are expected to arrive early next year. Meanwhile, although some faculties have already prepared their own guidelines, a uniform university policy on the use of AI is also being drafted.

gp