Dr Dominika Nadia Wojtczak

I recently completed my PhD in Computer Science (CDT Cyber Security) at the University of Bristol, where my research focused on applying artificial intelligence to understand and detect online misinformation. My work sits at the intersection of Large Language Models (LLMs), natural language processing, cognitive computing, and social media analytics.

My research developed a novel multimodal dataset combining social media posts, participant demographics, political orientation, social media behaviour, human annotations, and LLM-generated annotations to investigate how cognitive biases and personal characteristics influence the perception of misinformation. I also conducted one of the first comprehensive comparisons between human annotators and modern LLMs for cognitive misinformation analysis, evaluating their ability to detect contextual features such as misinformation, stance, sentiment, manipulative rhetoric, and sarcasm using zero-shot, few-shot, and chain-of-thought prompting strategies.

In addition, I explored how sentiment, hate speech, bot activity, user engagement, and visual characteristics contribute to the spread of misinformation across political, health, and societal discussions. My findings demonstrate that while LLMs can significantly improve the scalability of misinformation detection, human expertise remains essential for interpreting subtle, context-dependent, and ambiguous content. Overall, my research advocates for hybrid human-AI moderation systems that combine the efficiency of LLMs with human interpretative judgement.

My research interests include Large Language Models, generative AI, AI evaluation, misinformation detection, natural language processing, multimodal machine learning, human-AI collaboration, social network analysis, software engineering and trustworthy AI policies.