NER is explicitly useful for improving day-to-day life via automatic task creation. For example, when you ask your voice assistant to set an alarm tomorrow at 7am, or when you ask your smart home to dim your lights in the evening, you can be sure that NER was behind the scenes at work. Due to the vast array of potential situations here, a large amount of diverse data is required to train the algorithm properly.
In commercial uses it can be used in real-time to process social media discussions to provide insights into how people are viewing your brand right now, allowing you to react to it in real-time.
It is also used to predict demand by monitoring how people are talking about different topics, allowing e-commerce and finance platforms to predict velocity more accurately. The accuracy of the data used to train the systems has a direct impact on the accuracy of the systems themselves.
The final use case where we see NER frequently is when it comes to education and search. NER algorithms allow for a more accurate linking of information between different entities to make search results more precise. This requires again a large corpus of very accurate data.