ABSTRACT Donation‐based crowdfunding involves asking the public for financial help using persuasive narratives. Factors such as easy‐to‐understand messaging, campaign description length, perceived authenticity, and perceived deservingness influence fundraising success. As in other online practices, some crowdfunding campaigners are using generative artificial intelligence (GenAI) to write and edit campaign narratives and some crowdfunding platforms have integrated GenAI. In this exploratory study, we seek to understand how different GenAI platforms may alter crowdfunding narratives, whether these adjustments align with evidence on what improves crowdfunding success, and implications for how normative language is used in crowdfunding. We processed 93 crowdfunding campaigns from the GoFundMe crowdfunding platform through each of six GenAI models (Chat GPT, Gemini, Copilot, DeepSeek, Claude, and Grammarly) and two crowdfunding platforms with integrated GenAI (GoFundMe and AngeLink). We analyzed the resulting texts using natural language processing, sentiment analysis, and text statistics tools and examined how GenAI altered normative language. The median word count decreased or remained the same for all but one model. GenAI models tended to leave campaigns at a similar or higher readability level as the original text. Sentiment scores were more positive in all but one model. GenAI models tended to add normative language to the original campaigns. This study demonstrates that GenAI alterations are aligned with factors that are known to increase fundraising success. Stand‐alone GenAI models tended to add normative language rather than remove it and to focus on the individual desert of campaign recipients rather than the social conditions that gave rise to their needs.
Snyder et al. (Tue,) studied this question.