Lange language models (LLM) aggregate huge amount of data. As well they are able to generate new texts and interpret the provided text input. All these features are very common to humans and could be used to automate tutors work on e-course preparation and execution. However its usage is tricky, as the e-course must meet the course requirements. Specific solutions should exists to assure the LLM usage is adjusted and adapted to fulfill all e-course preparation and execution requirements. This challenge invites to develop those solutions and at least partly automate the tutors work for e-course preparation and implementation.
The direction for the challenge ideas, which could be converted into multiple final thesis or individual project topics is the following:
- Automated course content generation to meet course requirements using large language models and RAG framework.
- Automated course test question generation to meet the course syllabus, using large language models and RAG framework.
- Automated text answer correctness estimation, using large language models and RAG framework.
- Design and usability testing of user interface, adapted to increase user experience in automated or partly automated e-course environment.
- Prompt engineering for dialogue generation using AI
- Influence of prompt engineering on modern education
- Creating assignments using generative algorithms for the Bebras Informatics Thinking Competition
Regarding more details on the challenge and topics contact Simona Ramanauskaitė (topics 1-4), Antanas Čenys (topics 5-6) and Irma Šileikienė (topic 7), VILNIUS TECH.
Presented ideas (topics 1-4) are generated by Rujun Gao (Texas A&M University) and could have an additional mentorship from her side.