I actually think the move away from master/slave terminology is a good thing, not because of the association with human slavery, but because the term "master" is a bit too broad.
For example, in a database, the master is usually the running copy while the slave is the backup. First, it doesn't match the idea of a master and a slave, normally, the master commands and the slave works, a (human) slave doesn't sit idle and take the role of the master when he is not available. This is a backup, not a slave.
In the context of process execution, the parallel with slavery is more clear. The master process controls the slave processes completely while they do the work. Interestingly, these are often called "workers" even though it is the closest thing we have to slaves in computing.
In the context of a bus, the master can either simply be the #1 device, while the slave is #2, no real hierarchy here. In other cases the master is the controller and decides which device (sometimes called a slave) talks and which one listens. There is a clear hierarchical relationship but it is a bit different from the master/worker relationship of processes.
And now the "master" branch in git, which is actually more like a trunk (SVN term) or default (Mercurial term).
Rethinking the "master" term forces people to think of more precise terms, in most cases, "master" is just lazy.