The Minds, Machine, and Society group is led by Prof. Soroush Vosoughi, an assistant professor of Computer Science at Dartmouth. Underpinning our work is the profound interest in developing computational models of cognition, employing language as a primary conduit. We believe that the creative use of language is a distinguishing cognitive skill that sets humans apart from other animals. At a high level, our research primarily focuses on natural language processing (NLP) and machine learning. Specifically, we delve into the intricacies of large language models (LLMs) such as those exemplified by ChatGPT, in an effort to understand and mitigate their anti-social tendencies like promoting stereotypes, spreading toxicity, and misalignment with human values. To increase their transparency and trustworthiness, we work on developing a range of interpretability methods, breaking down their inherent "black box" nature. Our pursuits also encompass techniques such as reinforcement learning for guiding pre-trained LLMs and grounding them using simulations. We also seek to harness the power of LLMs and classical NLP to create computational tools that offer new perspectives on social systems and issues. Our work encompasses studying phenomena such as political polarization, bias, propaganda, rumors, and hate speech, falling under the umbrella of "computational social science". Our research horizons have recently broadened to include the integration of visual information into language models (i.e., visual-language models), aiming to create a more complex representation of the abundant, rich data that exists and drawing us closer to a deeper understanding of human cognition. Also, our exploration now extends to applying LLMs to the health and bioinformatics sectors, motivated by the intriguing parallels between genomic sequences and language. This venture is a collaborative effort with several other faculty at Dartmouth.
Prof. Vosoughi is a recipient of the Google Research Scholar Award in 2022 and an Amazon Research Award in 2019, and his work has earned several Best Paper awards and nominations, including the Outstanding Paper Award at AAAI 2021. He is a Technical Associate Director at Dartmouth's Center for Precision Health and Artificial Intelligence (CPHAI), a faculty member of the Institute for Security, Technology, and Society (ISTS) and a faculty affiliate of the Quantitative Social Science (QSS) program, the Quantitative Biomedical Sciences (QBS) program, and the Arthur L. Irving Institute for Energy & Society at Dartmouth. Before joining Dartmouth, Prof. Vosoughi was a postdoctoral associate at MIT and a fellow and later an affiliate at the Berkman Klein Center at Harvard University. During his time as a postdoc at MIT, he was the technical director of the "Electome" project. His team of graduate and undergraduate students developed a collection of tools for detecting and categorizing election-related content on Twitter, and they were invited to be an official partner of the Commission on Presidential Debates during the 2016 elections. In addition, Prof. Vosoughi served as a technical advisor to the nonprofit organizations Public Mind (formerly Because Humanity), a nonprofit using AI focused on developing AI systems that foster a healthier online public sphere, and Cortico, a nonprofit that builds systems for a healthier public dialogue, from 2021 to 2023 and 2016 to 2019, respectively. He received his Ph.D., MSc, and BSc from MIT in 2015, 2010, and 2008.
Prof. Vosoughi's research has been supported by a variety of prestigious entities. These include the National Science Foundation (NSF) and the National Institute of Health (NIH), alongside industry grants from renowned corporations like Google and Amazon. Furthermore, he has received funding from nonprofit organizations such as the Institute for Humane Studies and the John Templeton Foundation. His work has also benefited from internal grants at Dartmouth, awarded by the Neukom Institute for Computational Science, the Institute for Security, Technology, and Society (ISTS), the Dartmouth Ethics Institute, and the Arthur L. Irving Institute for Energy & Society.