Abstract Entrepreneurship research faces a growing challenge: studies using advanced AI methods are trusted less than those that do not. In a survey of 172 entrepreneurship scholars, we document a substantial trust discount—AI-based studies are perceived, as a category, to be significantly less credible than those using conventional statistical methods. To explain this outcome, we adapt Akerlof’s market-for-lemons framework and introduce the concept of “social lemons”: studies devalued due to deep epistemic opacity. This stems from a double-black-box challenge: opaque AI methods are applied to uncertain, elusive entrepreneurial phenomena. Echoing dysfunctional market dynamics in Akerlof’s market-for-lemons framework, this dual opacity can create conditions whereby flawed research goes undetected, while high-quality work is crowded out. We outline a multi-stakeholder roadmap for safeguarding research credibility, offering actionable measures to help transform today’s trust discount into a future trust dividend—positioning AI as an accelerator of insight, rather than a source of doubt.
Obschonka et al. (2026) studied this question.