# Dirk Hovy Dirk Hovy is a computational social scientist specializing in natural language processing and fairness in AI, known for his research on demographic biases in language models and algorithmic fairness. Dirk Hovy is a computational social scientist whose research focuses on natural language processing (NLP), machine learning, and fairness in artificial intelligence. He is particularly known for his work on demographic biases in language models and algorithmic fairness, examining how social attributes such as gender, age, and nationality are encoded in text and how these biases propagate through AI systems. Hovy has contributed to the development of methods for detecting and mitigating bias in NLP, and his work has influenced discussions on responsible AI development. Hovy is affiliated with Bocconi University in Milan, Italy, where he leads the Data and Marketing Insights unit and is a professor in the Department of Computing Sciences. He has also held positions at the University of Southern California and the University of Copenhagen. His research bridges computational methods and social science, addressing questions about how language reflects and shapes social structures. ## Early Life and Education Dirk Hovy was born in Germany and developed an early interest in both computer science and linguistics. He pursued his undergraduate studies in computational linguistics at the University of Tübingen, where he was exposed to the intersection of language and computation. He then moved to the United States for graduate studies, earning a PhD in computer science from the University of Southern California (USC). His doctoral research focused on statistical machine translation and the use of syntactic information in NLP models, laying the groundwork for his later work on social dimensions of language. ## Academic Career After completing his PhD, Hovy joined the University of Southern California's Information Sciences Institute as a research scientist. During this period, he collaborated with colleagues on projects involving multilingual NLP and the development of resources for low-resource languages. In 2015, he moved to the University of Copenhagen, where he became an associate professor and later a full professor. At Copenhagen, he founded the NLP section and led research on computational social science, including studies on the demographic correlates of language use. In 2021, Hovy joined Bocconi University as a full professor. There, he established a research group focused on fairness and transparency in AI, with a particular emphasis on NLP applications. He has also been involved in European research initiatives, including projects funded by the European Research Council, and has served on program committees for major NLP conferences such as ACL, EMNLP, and NAACL. ## Research Contributions Hovy's research has made significant contributions to understanding and mitigating bias in NLP systems. One of his notable works, "Learning Word Vectors for 157 Languages," introduced a multilingual word embedding model that has been widely used in cross-lingual NLP tasks. He has also published influential papers on the automatic inference of demographic attributes from text, demonstrating that models can predict a person's gender, age, and other traits from their writing style. This line of research raised important ethical questions about privacy and the potential for misuse of such predictions. In the area of fairness, Hovy has investigated how biases in training data lead to discriminatory outcomes in [machine learning](https://www.wikiprompt.org/wiki/machine-learning) models. His work has shown that word embeddings and [large language models](https://www.wikiprompt.org/wiki/large-language-model) often encode stereotypes, such as associating certain professions with specific genders. He has proposed methods for debiasing these models, including post-processing techniques and data augmentation strategies. His research has been published in top venues, including the Proceedings of the Association for Computational Linguistics and the Journal of Artificial Intelligence Research. ## Impact and Recognition Hovy's work has been widely cited and has influenced both academic research and industry practices. His findings on bias in NLP have been referenced in guidelines for responsible AI development by organizations such as [OpenAI](https://www.wikiprompt.org/wiki/openai) and [Google DeepMind](https://www.wikiprompt.org/wiki/google-deepmind). He has been invited to speak at conferences and workshops worldwide, including the Conference on Fairness, Accountability, and Transparency (FAccT) and the International Conference on Machine Learning (ICML). In 2020, Hovy received a prestigious grant from the European Research Council for his project on "Socially Aware Natural Language Processing," which aims to develop NLP models that are more sensitive to social context and less prone to bias. He has also been recognized with best paper awards at several conferences, including an honorable mention at EMNLP 2016 for his work on demographic inference. ## Selected Publications Hovy has authored over 100 peer-reviewed papers. Some of his most cited works include: - "Learning Word Vectors for 157 Languages" (with Anders Søgaard, 2017) - "Demographic Information from Text: A Survey" (with others, 2019) - "A Corpus for Sentence-Level Subjectivity Detection in German" (with others, 2013) - "The Social Impact of Natural Language Processing" (with others, 2021) His research has been supported by multiple grants from the National Science Foundation, the European Union, and private foundations. ## See Also - [Artificial intelligence](https://www.wikiprompt.org/wiki/artificial-intelligence) - [Natural language processing](https://www.wikiprompt.org/wiki/natural-language-processing) - Algorithmic fairness - Computational social science --- Source: https://www.wikiprompt.org/wiki/dirk-hovy License: CC BY-SA 4.0 (https://creativecommons.org/licenses/by-sa/4.0/) Last updated: 2026-09-05T14:09:40.593083+00:00