The UI may initially take some time, but not any more than learning how to use a pencil the first time. Glass feels unlike anything you've used before because it is. That said, my experience at Google I/O, I figured out how to use it very quickly and I thought the interface was intuitive as I navigated. Using the touchpad on Glass is more like using a smartphone before touchscreens. It feels a little disconnected in that it is indirect control, but when Blackberries only had a jog wheel, millions of users had no problem using those devices.
Given that Glass doesn't have a launcher like most Android devices, the real challenge will be in actively using Glass. Passive use, whereby you are connected to applications using the Mirror API, means that you just receive notifications pushed by remote services, and if you don't respond to a card, it just gets pushed down the stack. You can navigate backwards to find those cards again, but just as you don't respond to every Facebook post, you can just let them flush out and choose to ignore it.
The two problems that have to be resolved, are how do you control "native" applications, built using the GDK, and how do you control the flood of extraneous information when you might subscribe to tens or hundreds of services that send you messages through the Mirror API.
While you can control what notifications you receive, it would be very easy to subscribe to "important" services and still feel overwhelmed by notifications that aren't particularly interesting just so that you do receive the notifications that you do care about. Culling this stream into something more useful is where some of the AI research may also pay off. It is a difficult balance to figure out today, but as this tech evolves, I have no doubt this problem can be solved too.