Or, JSON is the new XML. Or both.
The thing is, for fast data transfer, we should all be using binary transfer protocols in a language rich with data structures, right? But no one (read: major play-ahs) could agree on such so XML was born as the independent data interchange protocol. Right, it's clunky and it doesn't (easily) afford the structuring of complex data structures, but it was a major advancement at getting legacy systems to talk with one another when they have to.
Who saw XML coming to the browser? Well, as a matter of fact, I knew a guy at the IAEA who had been using AJAX-like techniques five years ago, but it was an intranet website and he only had to worry about IE. I knew another guy there who was singing the praises of thin clients. His idea was that that HTML was too bulky (I mean, come on, it's markup and content all in one fell swoop, right) XML was the answer to have a minimal amount of content over the wire with the thin client being able to handle the markup/business logic. Sound familiar?
Looking at these JSON examples, I think the claim that "JSON is much easier for human to read than XML" is flat out wrong but I appreciate that for machines, it's all the same.