USI - Email

Archive / INF Seminars / INF_2025_04_04_Donya_Rooein

USI - Email

	Advancing Large Language Models for Children in Education: Adaptations, Biases, and Future Directions

Host: Prof. Monica Landoni

Thursday

03.04

USI East Campus, Room D0.03

11:30 - 12:30


	Donya Rooein Bocconi University

Abstract:
Large Language Models (LLMs) are rapidly transforming educational experiences by offering unprecedented opportunities to personalize learning, enhance student engagement, and support diverse educational needs. However, employing LLMs effectively for children introduces distinct challenges, including ensuring developmental appropriateness, mitigating biases, and addressing varied learning requirements. This talk emphasizes child-centric design in educational contexts to evaluate the performance and suitability of LLMs specifically for young learners. We critically examine the outputs of LLMs across different age groups and grade levels, assessing their appropriateness for supporting various educational objectives. Additionally, we explore biases embedded within LLM-generated content, investigating how such biases influence children's learning experiences and their perceptions of societal norms and values with different demographic groups.

Biography:
Donya Rooein is currently a Postdoctoral Fellow in the Department of Computer Science at Bocconi University, under the supervision of Prof. Dirk Hovy. Her research lies at the intersection of language, computation, and society, with a particular focus on demographics and education, all aimed at leveraging Natural Language Processing (NLP) for social good. She is passionate about applying her NLP expertise to drive meaningful societal impact. Recently, Donya visited the Language, Reasoning, and Education (LRE) lab at ETH Zurich, where she collaborated with Prof. Mrinmaya Sachan. Together, they co-organized a tutorial titled "Large Language Models for Education: Understanding the Needs of Stakeholders, Current Capabilities, and the Path Forward," which has been accepted for presentation at ACL 2025. In October 2022, Donya earned her Ph.D. in Information Technology Engineering from Politecnico di Milano, graduating summa cum laude under the supervision of Prof. Barbara Pernici and Prof. Paolo Paolini. Her doctoral research made significant contributions to developing a scalable framework that helps teachers integrate educational chatbots into their classrooms. Her Ph.D. was funded by a scholarship from EIT Digital (European Institute of Innovation and Technology) and IBM Italy.

Host: Prof. Monica Landoni