The difference is the scope of the project you want to do now, not 3 years down the line. How many solo UDK projects actually reach completion? - UDK is really more for teams of people, the media demands and internal knowledge, it's a helluva lot for 1 person to undertake.
The UDK system is great, it's probably as close to an ideal development environment as you can get, but why use a clawhammer to open a peanut. If you want to make something a bit retro, something fun to work on, then you should use the platform that is good enough to get the results you want. Personally, I flip between XCode, AppGameKit, DBPro, Purebasic, VB, VBA... the right tool for the job, it's how projects get completed.
What about AppGameKit? - frankly, I'd really like to see a less linear tablet RPG, less about selling you gear, more about just enjoying a game. Seriously, some of the RPG's on mobile devices are pathetic. A survival horror game in the same style as Amnesia for instance, with some RPG elements, well I'd buy it. Mobile games tend to be about little portions of gameplay, bite sized chunks, even though there's a solid market for people who like these games. Look at Enderman, some trees, some empty, basic buildings, 1 bad model, vague lighting - and it's hugely popular!. Enderman could have been written in DBC, never mind DBPro, never mind GDK. I mean, if you like to concentrate on art, surely it makes sense to keep concentrating on art, because AppGameKit doesn't take much learning at all. Make an artistic game that let's you adhere to your talents and is still fun to develop.

I got a fever, and the only prescription, is more memes.
