Depends on the definition of “improve”. No doubt it will gain access to greater volumes of information, and be taught to interact with the public is a more “human like” manner. This will not necessarily coincide with a greater understand on the part of the programmers in moving AI to a safer, less threatening to humans condition. Developer competition is fierce. Follow the money. It will dictate the priorities the competitors require of their systems. Safety for the public, reliability of product, and long term discipline of application will likely be overcome by - guess what? - $$$$$. Greed tends to be strongly in control of commerce. There seems to me to be a probability of great difficulty in the field of AI in our future, and not far out in time. The difficulty will not be confined to AI producers, but to those who implement that technology across the entire breadth of our society.
I am not a doom sayer, but I am one to caution we are now at an inflection point where some reasonable limits and controls must begin to be implemented by the federal government of the US in cooperation with all major governments. I ordinarily am not a fan of governmental control of private industry, but AI is somewhat like atomic power. I need not comment further.