Can LLMs Change Our Past?

This talk presents results of an international, interdisciplinary collaboration of memory studies scholars, philosophers, and computer scientists, assessing the impact of large language models (LLMs) on the formation of collective memory. In traditional terms, collective memory is a dynamic product of data selectively provided mainly from the institutions of memory (e.g., archives, museums, media, schools). However, AI changes how we access and use data in public spaces. This project's main research focuses on the social and educational uses of AI, how AI reshapes the process of declarative memory-making, and how it influences an ethically sustainable and impartial social framework for the construction of collective memory. Practically, how does our use of LLMs shape our collective memory concerning critical historical events? We analyzed several historical conflict events across languages, gaining important insights to face the practical and ethical implications of the growing use of LLMs for education, as well as for digital policymaking and peace-building.