The PKIM is designed to facilitate high-speed input of natural language text using only a basic phone keypad. It relies on the correspondence of each phone key to a set of letters from the natural langauage alphabet (eg. the "2" key might represent "A", "B", or "C"). For many languages, most words in the lexicon are uniquely specified by the correspondence; for example, with a typical phone key letter correspondence, a test lexicon of 45378 English words had only 2942 (6.5%) that were ambiguous (eg. the sequence "8439" corresponds to the set of ambiguous words "they", "view", and "tidy").
For each set of ambiguous words, we can choose in advance which word is more likely (eg. in a sample of text, the word "they" occurred 8 times more often than "view", and 20 times more often than "tidy"). In a sample text containing 3438671 words, less than 1% where one of the less-likely words in an ambiguous set.
If the user mainly views the keypad while entering text, then possibly dire communication mistakes might occur where "Davies has pooled our funds" becomes "Father has smoked our fumes". Typos produced by PKIM are less likely to be obvious to the recipient that those produced on a regular keyboard. In this case, there seems a strong need to force the user to check such ambiguities.
If the user mainly views the screen while entering text, errors are non-critical, and it is more important to speed the flow of communication, and hence it is better to silently select the most common choice of ambiguity and allowing them to see and change this in 1% of the words they enter.
Unfortunately, operating the PKIM while looking at the screen is a skill that new users will take quite some time to learn.