(I have a list of six predictions about what I expect to see happen in academic library technology over the next 5-10 years. This is prediction #3.)
Visual search will (at a minimum) reach parity with text-based search.
One reason I have this prediction is because of what I believe is going on with mobile devices. More stuff on mobile devices + more mobile devices with touch screens = greater questioning of why we should force people to tap-tap-tap their way through text-based interfaces. "Im in ur txt 2 say luv u 4ever" and its ilk evolved for a reason, and while a lolspeak subject thesaurus would be amusing, I'm not counting on it.
A second reason is the increasing sophistication of infographics. (Digression: in reading this post by Rob Paterson, I was struck by how one cites information from an infographic--by x and y coordinates?) Designs such as Anil Dash describes in "Pixels are the New Pies" lend themselves quite well to the size and shape of mobile device screens, too. It seems like EBSCO's visual search is working off of similar design principles. Good for them.
The third reason is things like bing's image search. This kind of technology is also on my June list to Learn What I Am Talking About. Not only because users will increasingly like these kinds of tools, but because content will more and more be born multimedia. The Journal of Visualized Experiments will have many, many fellow publications in all fields before too very long. Say, 5-10 years.