What is the future of NLP?

As the new models received more attention both in industry and academic, most companies have been headed to invest their budget on hiring new talents. I remember when I was in graduate school in 2010, thinking on Natural Language Processing in my mind was limited to only information retrieval systems that retrieves records through TF-IDF methods. In last perhaps 5 years with the power of more advance GPUs a new evolution of data happened and this was over unstructured text. Deep Learning and Text Mining became two siblings that couldn't live without each other; more research papers published in conferences and journals.

Google and Facebook, two giant companies released their latest achievements in form

of pre-trained models that were trained over very large corpuses.

Bidirectional Encoder Representations from Transformers - BERT, is

a deep learning model that holds the context of a sentence and calculates a vector representation of a each single word in a sentence as a given input. Now, Researchers don't need to run their own model to train over a corpus and can leverage the pre-trained models for some down streams tasks such as text classification or sentimental analysis. After BERT was born, new other extended versions of it created like RoBERTa or AlBERT. And this became just a beginning.

I can see in the next 5 years NLP will become an essential ingredients for almost any products that has related to high volume of textual format. Products that are in proof of concept phase today will be come into markets either as individual form or through web services. Chatbots, Intelligent Assistants, IVR systems, Semantic Search are going to become sophisticated to not only interact with users but also present empathy, emotion, sentiment and perhaps even deeper understanding with them.

Would this future be better or worse, What do you think?