Studying Bias in the Age of AI

As artificial intelligence systems increasingly mediate access to information, understanding the interaction between human and algorithmic bias has become a central societal challenge. Search engines (SEs) and large language models (LLMs) are often perceived as neutral intermediaries, yet their outputs are shaped by training data, optimization objectives, platform incentives, and user behavior. Through examples and drawing on a case study conducted before the 2024 European Parliament elections, this talk will explore how human and algorithmic biases interact and reinforce one another, contributing to fragmented informational realities. In particular, it will present a privacy-preserving bot system that was used to perform synchronized searches across locations and query types, revealing consistent amplification of particular political entities and narratives by both SEs and LLMs.