The landscape of driving automation is dynamically evolving and driving automation gets more and more poreful. However, despite promising advances, there are still numerous shortcomings that need to be addressed to ensure safety, efficiency, and user satisfaction. Public discourse, especially in social forums, social media, and other feedback platforms, is rich with user experiences and insights that can be instrumental in understanding the current gaps in driving automation systems.

The primary goal of this thesis is to leverage the capabilities of Large Language Models (LLMs) to encode and analyze large amounts of textual data from various social platforms to generate a nuanced understanding of the public’s perceptions and experiences regarding the shortcomings of existing driving automation features.