Artificial Intelligence (AI), and specifically large language model AI, presents a wealth of opportunities to advance entrepreneurship scholarship. One of which is exploring ChatGPT as a new type of methodology to advance research on social entrepreneurial rhetoric creation, framing, and refinement. Utilizing prompt engineering as experiments, we engaged ChatGPT to produce a corpus of social enterprise ideas (n=311) aimed at alleviating poverty in Africa. We then tasked ChatGPT to refine these ideas (baseline corpus) into persuasive pitches in a ‘pitch to win’ scenario (pitch corpus). By comparing the baseline and refined pitch corpus using a mixed-method that combines natural language processing analysis and qualitative interviews, our analysis unveiled that ChatGPT strategically employed a variety of entrepreneurial framings in the language refinement process. At the word level, it leveraged call-to-action, impact, visionary, and collective framings. At the phrase, sentence, and paragraph levels, it employed solution, reinforcement, and plotline framings. In addition, by viewing entrepreneurial framing as an external, meaning-evoking strategy, we theorized a process model that demonstrates how entrepreneurial framing leads to desired communication outcomes by engaging audiences in a meaning-making process. Methodologically, our research reports one of the first empirical evidence of the potential of AI in refining social entrepreneurial rhetoric through a ‘prompting as experimenting’ approach. This work paves the way for future research to explore the limitless possibilities of AI for social entrepreneurship research, practice and policy with a particular attention on rhetoric.
