I would throw in some dead ends along the way as they very much altered the trajectory. LISP was the be all and end all in "AI" for quite some time as an example.

Many of the things being mentioned in these comments are how we got to where we are, but lots of time was spent on now dead evolutionary branches. Not a whole lot of practical progress in the 80s. I remember chatting with "AI" researchers in the 80s and they were on and on about Hebbian Synapses.