The NYT has a blog post by Virginia Heffernan about adaptive keyboards and interfaces. The summary: "Having touch-screen displays that look like keys but are really responsive mini-apps introduces excitement but also unease."
The article is kind of a "what if" piece about how adaptive software, including keypads, could destabilize the experience for people. Link: The Promise and Peril of 'Smart' Keyboards.
The text-input examples given (from Apple's iOS devices and Swype) are really only the tip of the iceberg when it comes to shaking up qwerty. There's a rich history of research into text-input methods, with many more radical, and in some cases more efficient, ideas -- look at Dasher from the Cambridge University for example. For a serious book-length treatment I'd recommend Text Entry Systems by MacKenzie and Tanaka-Ishii.
Related: The BBC recently had a good write-up on the history of Qwerty: Why do we all use Qwerty keyboards?