I think you missed the point. The point isn't chat == text-only, it's that chat == natural-language-style UX. This could well be augmented with images (like is with Siri). Basically "is there a button to do what I want" versus "do I indicate what I want in a human-ish way" - imho that's the defining feature of a "chatbot", not that it's text-driven per se.