More parts of systems have a property where you want to put more data into it and it gets better. This has led to a convergence towards thinking of problems as a fusion problem in machine learning.