AI-simulated entrepreneurship under uncertainty: forecasting university-driven capability evolution

Abstract Universities represent the crucial nexus between research and technology transfer, yet the high venture failure rates raise a fundamental question: Are academic institutions failing to equip entrepreneurs with the capabilities essential for navigating uncertainty? Despite expanding entrepreneurial programs, universities maintain outdated knowledge-delivery models focused primarily on traditional horizontal and vertical business knowledge, rather than crisis-relevant capabilities. This study examines whether universities develop essential entrepreneurial capabilities for navigating uncertainties, such as psychological and community-related, by investigating the manifestation of four capability domains: horizontal, vertical, psychological and community-related. By developing and training an AI-bot investor simulation, we evaluated how academically-supported versus independent entrepreneurs signal these capabilities and attract investor interest. Using LLM-based topic modeling and sentiment analysis, we discovered investor assessments strongly favor psychological and community capabilities alongside traditional business expertise, with substantial value placed on positive sentiment across all domains. This preference creates a critical mismatch with academically-supported entrepreneurs, who disproportionately emphasize vertical knowledge while neglecting psychological and community domains—resulting in lower overall positive sentiment that undermines their investment appeal. Surprisingly, while academic affiliation itself provides inherent credibility with investors, universities paradoxically fail to capitalize on this advantage. This missed opportunity becomes especially compelling as our data shows correlation between discussion frequency and positivity, especially in psychological and community domains. This finding represents fertile ground where university programs could refine entrepreneurial preparation. Our research advances effectuation theory and the CAVE model and challenges dynamic capabilities (DC) by demonstrating how complementary capabilities outside university’s traditional focus impact entrepreneurial navigation in uncertainty.