Voice assistants such as Alexa continue to improve in their capabilities and utility for customers. Since launch, the use of voice as an interaction modality has expanded, and talking to such assistants is ever more natural. One of the next steps in voice AI becoming a truly ambient intelligent service is in the development of increased reasoning and generalised intelligence technologies, both to complete more complex tasks, but also understand the latent goal of a user. Even what can appear as a simple interaction on the surface can require a lot of knowledge and reasoning capability to understand and act. In this talk we will use several examples to discuss a potential evolution of reasoning in voice AI.
Craig Saunders is Director of Machine Learning in Alexa AI. He leads the Evi team, which was one of the early acquisitions that contributed to Alexa’s technology. Craig completed a PhD in 2000 in Machine Learning and spent 9 years as a University academic before moving to industry at Xerox Research Centre Europe. He has conducted fundamental research into machine learning, NLP and computer vision, and applied research to deliver state of the art solutions in many industries including finance, legal, healthcare, and BPO processing. Craig joined Amazon and the Alexa team in 2017, where he’s worked on scaling Alexa’s knowledge and question answering capabilities.
