Raphael Hernandes

PhD thesis: TBC

Research interests:

1. AI Ethics & Algorithmic Auditing
2. Algorithmic Bias & Media Diversity
3. Epistemic Security
4. Generative-AI in Journalism

My PhD will analyse the impacts of the wide-scale deployment of generative web search through Large Language Models (LLMs) on the distribution of news. Such Artificial Intelligence (AI) systems are being deployed both as standalone products, such as Perplexity, and embedded within traditional search engines, like Google, providing search results as AI-generated text. This raises critical concerns about potential biases and the distortion of information, which could undermine democratic processes by affecting public knowledge and opinion. My research aims to create a novel methodology to analyse these systems and their potential impacts in news distribution based on algorithm auditing techniques to spot distortions in source selection, framing, and accuracy. Building on that analysis, it will propose interventions to safely deploy these systems and actionable recommendations for platforms, regulators, and newsrooms.

Who or what inspired you to pursue your research interests?

Web search is the main channel through which users access news websites today. These systems are being replaced by generative AI, meaning generative search is fast becoming the front door to the news around the world. Because their ranking logic is proprietary and the content is personalized to the users, society cannot see when whole viewpoints are biased, disinformation creeps in, or when specific news sources are privileged over others as sources of information. Yet these systems have quietly inherited the gatekeeping role once held by editors. Democratic health depends on plural, accurate information; scrutinising and fixing these opaque filters is therefore urgent. As a journalist who has reported for Folha de S.Paulo and The Guardian, I feel a professional duty to make these systems legible and fair to the publics they now shape.