April 11th, 4:30 PM
CHI Think Tank, Frost Library
Abstract: Artificial intelligence (AI) and natural language processing (NLP) are coming to prominence with startling new advances, but their functioning and potential is poorly understood, with incredibly weak scientific grounding of every cool recent result you’ve heard of. Why?
Considering the history of AI research, its major approaches are sometimes conceived as a “neats versus scruffies” dichotomy: formal mathematics (e.g. logic) versus ad-hoc engineering (e.g. neural nets). But both views are incomplete. AI is better viewed as a social science: all successful computational models of language and human intelligence are incredibly bound to the complexities of the human experience, from our behavior to cognition to social institutions. In the first part of my talk, I’ll argue this helps us understand controversial trends in recent deep learning advances in vision and language modeling - for example, models’ deep dependence on human behavioral data, their resulting internal cultural and social biases, and constraints resulting from the capitalist and military institutional contexts of their development.
But by acknowledging the deep social embeddedness of AI technology, we can better map out research directions with better social awareness and insight. In the second part of my talk, I’ll overview NLP-for-social-science work where we utilize natural language processing for news and social media analysis, to gain deeper understanding of social scientific topics such as racial dialects in the U.S., or political violence in India. Applying NLP as computational social science - while being aware of its limitations and social confounds - has enormously promising potential with ongoing advances in these tools and methods.
Brendan O’Connor
Speaker: Brendan O’Connor is an associate professor in the College of Information and Computer Sciences at the University of Massachusetts Amherst, who works in the intersection of computational social science and natural language processing. At UMass, he is an Associate Director of the Computational Social Science Institute. He holds a PhD in Machine Learning from Carnegie Mellon University and BS/MS in Symbolic Systems from Stanford University; he has previously been a Visiting Fellow at the Harvard Institute for Quantitative Social Science, and worked at technology companies including Facebook Data Science and Crowdflower
