XML gets people who have no business doing serialization and parsing out of that business entirely. Instead, we have schemas (either purely ad-hoc, or formalized) and common libraries to express those needs. Like any computing tool, XML has been used for good, bad, and ugly ends... but it brought the idea of schema-based document format and consistent Document Object Model (DOM) APIs to the forefront of data/document handling. In essence, this is a two-level interface: the document schema (a first-class entity, unlike an ad-hoc data format) and the programming APIs.
Even if you're no big fan of XML, the good ideas, patterns, anti-patterns, and mindset changes that spread along with it have positively influenced many other useful document technologies: JSON, YAML, the newer binary serialization formats (such as Protocol Buffers, Thrift, and Avro), and more.