Abstract: Natural Language Processing (NLP) has seen impressive gains in recent years, being integrated into useful technologies with improved capabilities, measured in terms of how well they match human behavior captured in web-scale language data or through annotations. However, human behavior is inherently shaped by the cultural contexts humans are embedded in, the values and beliefs they hold, and the social practices they follow. On the other hand, these technologies are often developed within mono-cultural development contexts, but are meant to interact with multi-cultural usage contexts with divergent values, knowledge systems, and interpretive practices. In this talk, we use culture as a lens on responsible NLP, and demonstrate how it is crucially influenced by culture along three dimensions: culture as reflected in language data, cultural values and norms that we encode in our models, and cultural knowledge systems we rely on for evaluation of our models. More concretely, we demonstrate how geo-cultural differences shape what language is deemed abusive, stereotypical, or offensive, and where NLP models fail in detecting them.
Bio: Vinodkumar is a Research Scientist at Google, working on issues around Ethical AI and ML Fairness. Prior to this, he was a postdoctoral researcher in the Computer Science department at Stanford University, where he worked with Prof. Dan Jurafsky and others at the Stanford NLP group, in an array of projects with a focus on applying Artificial Intelligence for Social Good. He brings together natural language processing techniques, machine learning algorithms, and social science methods to build scalable ways to identify and address large-scale societal issues such as racial disparities in policing, workplace incivility, gender bias and stereotypes, and abusive behavior online. To this end, Vinodkumar collaborates with researchers in Machine Learning and NLP, as well as social scientists and practitioners from Linguistics, Psychology, Education, and Journalism.