happens quite a lot in my experience
In particular, the UI often is non-static and can be in certain states, with the effect of user actions (and possibility of certain actions) depending on that state. If it is implemented or "grown" ad-hoc, it sometimes may turn into a jumble of interdependent code and flag/visibility variables, but very often the exact same behaviour can be modelled and implemented much simpler if you think of it as a DFA.
For example, the UI of recently popular single-page webapps, where JS is used to change 'screens' without page load, or change functionality by hiding/adding large parts of the page, can often be naturally be thought of as DFA, planned (on paper or wireframes) as states='screens' and transitions between them, and then implemented as DFA.
Thinking of UI as a DFA also forces you to consider the 'proper' effect of every user action in every state, which is just good UX practice. For example, in a classic url-based web app you may draw a graph/process of how an user proceeds through making a purchase on your site. Usually it will include only the typical process - but if you think of it as a DFA, you define behavior for actions such as 'requested url /ordervalidation/step2-address' at every state, since the result likely should be different depending on if the user went there from step1 by clicking 'ok, i want to buy this', step3 by clicking the back button, or two weeks later by clicking on a link that he accidentally bookmarked.