AlwaysOn Availability Groups, first introduced in SQL Server 2012 is a feature that is conceptually similar to database mirroring. I’m going to assume you already know what AlwaysOn Availability Groups are in general, their main purpose and how they are different (essential aspects at least) from other technologies to provide replication such as Database Mirroring.
In SQL Server 2014, the significant enhancement to availability groups was the increase in the number of supported secondary replicas from three to eight. SQL Server 2016 includes a number of new enhancements
The biggest thing about it, as I recall, was stability: I wouldn’t have recommended too many places go into production with Availability Groups in 2012, but by 2016, many of the biggest bugs were ironed out.