Chat interfaces are not the default. They are one option. UX teams have collapsed a full toolkit of input and output modalities into a single text box because LLMs are trained on dialogue, and that architectural fact has been mistaken for a design mandate. The result is products that fail users at the moment of highest need: a traveler sprinting through an airport who must stop, balance a coffee, and type a booking reference into a chat prompt, then parse a paragraph of weather explanation to find a buried gate number.
The article builds a practical case against modality laziness using two mechanisms: a Task Audit and an Input/Output Alignment Matrix. The input problem is framed as a linguistic barrier. A blank chat box forces users to become prompt writers, translating vague intent into precise language. A designer who knows exactly what lighting they want cannot necessarily describe it in text. A slider can. The output problem is framed as a cognitive tax. Text is a serial medium. A doctor needs a numerical display for vital signs, not a narrative. A stock trader needs a line graph, not a written description of price movement. Defaulting to text in both directions pushes interpretive work onto users who cannot afford it.
The full piece includes a taxonomy of input and output modalities mapped to the contexts where each performs best, plus an accessibility argument that modality design should multiply pathways rather than consolidate them. The Task Audit and Alignment Matrix are the tools worth reading for. They give teams a repeatable process for evaluating physical load, cognitive load, and immediate user intent before a single UI decision is made.
[READ ORIGINAL →]