Alexander Hoyle is a Researcher at the ETH Zürich AI Center, concurrently contributing to natural language processing/machine learning and social science groups. He holds a PhD in Computer Science from the University of Maryland (advised by Philip Resnik) and a Master's in Computational Statistics from University College London (advised by Sebastian Riedel and Jeff Mitchell). His research focuses on computational social science, emphasizing methods for latent construct identification (e.g., topic models, ideal point models) and evaluation frameworks grounded in validity. Key areas include bias/fairness in AI, political science applications, and mental health constructs like suicidality. He pioneered frameworks like PairScale (attitude measurement via pairwise comparisons) and TopicGPT (prompt-based topic modeling). Education: PhD in Computer Science, University of Maryland (2020-2023) MS in Computational Statistics & Machine Learning, University College London (2018) Bachelor's degree (pre-PhD details omitted) Research Interests: Combining NLP with social science needs, particularly in evaluation rigor and interpretability. Active in interdisciplinary work between NLP and computational social science (e.g., measuring attitude evolution on Reddit, improving topic model validity). Advocates for human-in-the-loop approaches to address LLM limitations in tasks like document clustering and sentiment analysis. Grants & Projects: Contributed to a landmark $2.2B DOJ settlement on NYC public housing via econometric modeling at The Brattle Group. Active in graduate labor advocacy (Maryland state legislature testimony) and mentorship (Científico Latino's mentorship program). Labs & Teams: Leads initiatives at the ETH Zürich AI Center, collaborating with groups like Microsoft Research (FATE) and AI2's AllenNLP. Involved in multi-university projects (e.g., University of Maryland's Computational Linguistics lab).







