We also developed a framework to break down the articulation of an assumption as premises to a target. Our framework simplifies the articulation process for practitioners in communicating their argument and supporting them in deciding future courses of action.

Our interviews with practitioners deconstruct vague conceptualizations, inconsistent views, and how an assumption traverses the phases of a typical AI workflow and leads to downstream effects in data pipelines and modeling while possibly begetting more uncertainties.

We discuss this marginalization of assumptions through the lens of argumentation theory from Informal Logic and frame assumptions as the **premises of an argument** an individual makes or uses to achieve a goal.

Practitioners are often required to list or explain their assumptions, but ironically, assumptions are often **assumed** to be straightforward to conceptualize, articulate, and interpret.