InferSent

I'm cofounder of Ought , a non-profit doing research on using machine learning to support deliberation. Previously, I was a researcher in Noah Goodman's Computation & Cognition lab at Stanford. And before that, I was a Ph.D. student in the Department of Brain and Cognitive Sciences at MIT, working in Josh Tenenbaum's Computational Cognitive Science... group . The question that drives my research is: How can we better solve our problems? Machines are potentially much better at answering hard questions if such questions can be posed formally. For example, given a formal model that describes how diseases cause symptoms and a formalized prior belief about disease rates, machines can reason precisely about what observed symptoms imply. In its various aspects, this question touches on mathematics, computer science, psychology, and philosophy: What does a language look like that makes it easy to formalize our prior knowledge and that allows us to pose the questions we want to answer? Given a formal problem statement in such a language, what do algorithms look like that can efficiently find answers? How is knowledge structured and used in the human mind, and can this inform our approach to the first two questions? Any particular question derives from the more general question What should be done? How can we understand more precisely what this question means? Machine Learning (ML) has been successful in automating a range of cognitive tasks that humans solve effortlessly and quickly. Yet many realworld tasks are difficult and slow: people solve them by an extended process that involves analytical reasoning, gathering external information, and discussing with collaborators. Examples include medical advice, judging a criminal trial, and providing personalized recommendations for rich content such as books or academic papers. There is great demand for automating tasks that require deliberative judgment. Current ML approaches can be unreliable: this is partly because such tasks are intrinsically difficult (even AI-complete) ...Read More