tl;dr: A better interface or a much prettier game would not turn me away in the abstract, but the reality of the situation is that the only way the second would happen would be to cripple things in such a way that *would* turn me away.
These are two separate affairs. Personally, I think the user interface could use some improvements, particularly in terms of consistency; the secondary-scroll mechanic in particular is problematic, and the game is generally bad about explaining why you can't do something. Being reasonably competent at operating (not playing) DF requires use of the wiki, the forums, or a lot of fiddly trial and error... the difficulty should come from the game and the world, not the interface. This is not, strictly speaking, a complaint; it's a game in early alpha with many years left before release, and a certain degree of clunkieness can be expected.
On the flip side, I find that tiny, bad art does a worse job of being usable than the pseudo-extended-ASCII art. I do usually swap to a 12x12 or 16x16 square set on larger monitors, but in general when you need to look over a large area with a lot of detail, icons do better then pictures. (One of the really interesting things about the Stardock "Elemental" games is that they are 3D representational when zoomed in, but go to a "cloth map" version with clear icons when you zoom out far enough (configurable). If DF ever gets a full 3D engine, this would seem to be the way to go.)
I'm also in the position to remember the huge drop in game quality when the PS2 came out, and so many game lines went from beautiful 2D to terrible 3D art that took longer and was more expensive. It took many years, and frankly the next generation of consoles, to get back to the level of clarity, usability, and craftsmanship / artistry we had.
Most importantly, however, is that DF is fundamentally about being a meta-generator for *many* different types of worlds. Trying to create a procedural art system to adequately represent, let alone artistically represent, the wonder and theme of a procedurally-generated mythos and culture is a task probably beyond even an AAA studio at the current tech level. I'm not saying it will always be impossible; but what we need to be thinking about in these early alpha days is how to best represent more abstract concepts. In the future, we will have worlds with few to no creatures pre-defined; we need to be not thinking about how to best draw an elephant in the limited pixels we have, but what icon to represent a very large herbivore that is passive by default but dangerous if provoked, because whatever roams the savanna -analogue may have no visual similarities to an elephant at all.
What we really need is a "fantasy" version of the
NATO symbols. They have been refined over more than 30 years, and are specifically designed such that even if you have no idea what the specific thing being represented *is*, you have a general idea of its relevance and importance to the overall tactical or strategic picture from the way the icon is constructed from meaningful parts. One might combine symbol sub-elements for ground-based, predator, large size; with additional elements for "can leap" and "can stalk / ambush" to get the icon for a panthera (great cat) like creature, for instance. But if on another world that ecological niche is filled by some weird lesser drake / lizardy thing, it might have the same *functional* icon, even if it's a leathery purple thing instead of a furry orange thing.