Evaluating Beyond the Screen: Collective Assessment of AI-Generated Business Plans with Resource-Constrained Entrepreneurs

Abstract

Entrepreneurs increasingly use end-user generative AI technologies such as ChatGPT for high-stakes documents like loan applications and business plans, where AI-generated errors---a wrong price, a fabricated product---can affect loan or funding outcomes. Current approaches to supporting evaluation of AI-generated text assume a single user assessing output alone, on screen. This can be especially demanding for resource-constrained entrepreneurs, whose digital and AI skills vary widely. In this early-stage work, we explore how evaluation might instead be organized in a group setting and completed as a collective activity. We extended BizChat, an AI-powered business-planning tool, with an evaluation module that links each generated claim to the entrepreneur's original input. We partner with community organizations in Maryland---embedding BizChat within various entrepreneurship programs---where workshop attendees (N=14) evaluated their plans through think-pair-share discussion. Early findings suggest interface scaffolds like claim-to-input links primed attendees with concrete, personal evaluations, which the group setting then extended beyond the screen: attendees requested printed copies, used rubrics to compare across plans, and drew on peers' knowledge to verify what they could not easily judge alone.