History a la ChatGPT

In this workshop we will collaboratively examine how different LLMs represent controversial events in recent history, such as the genocide in Palestine or the case of Afghan women under the ruling of the Taliban. We will test how easy or difficult it is to make them generate factually inaccurate and ideologically biased narratives. As internet users have increasingly started to search not for sources but for information, we will hopefully gain some valuable insight into biases that might interfere with LLM mediation of information. Also, we will see to what degree LLM providers have attempted to safeguard models from malevolent usage.

Hosted by: Johannes Widegren, Nasrin Mostofian and Mo von Bychelberg