Researchers at OpenAI have developed a method for entity disambiguation using a neural network that categorizes words into around 100 automatically-discovered 'types'. These types are non-exclusive categories, meaning a word can belong to multiple types. The system is designed to determine which object is meant by a word in a given context. --- Why it matters: This matters to AI engineers because it's an important step towards improving the accuracy of natural language processing tasks such as entity recognition and question answering. Source: https://openai.com/index/discovering-types-for-entity-disambiguation