FOCUS SESSION: AI is a game changer. AI is built on data. Half of that data is textual. Textual data is always multilingual. Unfortunately, most of us are excluded from the benefits of AI applications, since they only support English or Chinese. AI vendors typically claim their systems could be trained with any language, but this stays too often a theoretical possibility. We will discuss with leading AI executives how to architecture AI applications so that multilingual does not become an expensive (or impossible) afterthought. Language-agnostic design patterns are presented, which make sure that efforts for crucial tools such as chatbots, natural language generation, or knowledge extraction are not multiplied by the number of support languages. Case studies show the power and tremendous opportunities in multilingual AI.
