Traditionally, entrepreneurs had to engage in creative ideation, generate a couple of ideas, and consequently evaluate their opportunity status. Yet, as we argue in our paper published in the Journal of Management Studies (here), the emergence of Generative AI (GenAI) dramatically transforms this landscape. Novel entrepreneurial ideas are no longer a scarce resource, and any creativity gaps are being bridged. Although GenAI eradicates creativity bottlenecks, it simultaneously brings another problem to the forefront: entrepreneurs now confront an ever-expanding universe of conceivable entrepreneurial ventures that may nevertheless not stand for real opportunities. In effect, although machine imagination may bolster the creative reach of economic agents, it simultaneously radicalizes uncertainty. As we argue, GenAI reshapes the business environment in a manner that Knightian Uncertainty transforms into a “grand epistemological challenge”.
A grand epistemological challenge
The nature of this novel challenge does not lie only in the quantitative proliferation of technologically-assisted ideas. More critically, this challenge rests on the fact that knowing with certainty whether ideas represent opportunities is feasible only ex post; that is, the moment that the end-states driving entrepreneurial deeds do actualize. In addition, the ever-expanding universe of ideas generated by machine imagination is crowded by “false positives”; that is, nonopportunities misrepresented as opportunities. This is a problem because the opportunity cost is not only associated with the failure to identify a real opportunity. More importantly, embarking on “nonopportunity paths” can entail considerable personal and social costs.
For example, consider an entrepreneur who uses GenAI to identify a market gap for the development of a new health supplement. The analysis generated by the GenAI system looks compelling: market-data, consumer trends, and competitors’ positioning all look to align. The entrepreneur spends savings and one year to develop and launch the product, only to discover that the apparent opportunity never existed but was a figment of machine imagination. Besides the loss of money and time, and entrepreneur also lost the opportunity to pursue more realistic ventures. Alas, extant theory is somewhat fatalistic about the possibility of ex ante distinguishing opportunities from nonopportunities, thus fails to offer practical advice in the face of a world of radicalized uncertainty.
Beyond fatalistic views of uncertainty
Having revisited foundational assumptions and conceptualized the shape of the “grand epistemological challenge”, this paper moves to identify constructive ways out of it. To this end, we urge the need to move beyond fatalistic ways of thinking about knowledge, according to which the lack of perfect forms of knowledge means that we cannot know how to rationally tackle the opportunity/nonopportunity dilemma. We identify a more promising way out of the current impasse in “ecological rationality” – a research programme associated with the paradigm-shifting work of Gerd Gigerenzer. From a Gigerenzerian standpoint, we should move beyond idealized conceptions of knowledge and try to rationally navigate irreducibly uncertain environments through heuristics. Those are rules of thumb that can rationally guide decision-making when tailored to insights about the nature of the decision environment. For entrepreneurs and organizations, this means that although they can never attain infallible knowledge of whether an idea represents an opportunity, they are not condemned to act “in the dark”. Heuristics attuned to the structure of their decision environment can offer sensible – if imperfect – guidance.
The ECR model of opportunity search
We accordingly devise a three-stage ecologically rational model of opportunity search, viz., the Expansion, Contraction, Realism (ECR) model. The Expansion stage aims to ensure that real opportunities will populate the decision space through a targeted expansion of candidate ideas. The Contraction stage facilitates the emergence of “winners” through the structured elimination of nonopportunities. The gist of this most critical part of the model is that we ought to move beyond the traditional focus of opportunity-validation to processes of nonopportunity elimination. Last, the Realism component explains why the elimination task should be led by entrepreneurs: Whereas artificial intelligence excels at creative ideation, human intelligence equips entrepreneurs with a more realistic sense of possibility. We also suggest that the ECR model can be tested empirically through “live experiments” made possible by machine learning technologies.
Rethinking entrepreneurship and entrepreneurs
Whether this particular model survives empirical scrutiny is secondary to our higher-level motivation; namely, to advance theoretically rigorous ways out of fatalistic treatments of Knightian Uncertainty. As we argue, we can no longer afford this standard academic stance toward uncertainty in an era in which GenAI is making creativity abundant at the cost of the radicalization of uncertainty.
As importantly, our theorization does not only help us revisit standard assumptions about entrepreneurial creativity and opportunity search strategies. It also facilitates breaking ties with standard notions of entrepreneurship as the effect of a “rare breed” of economic actors – the supposedly extraordinarily creative force of the economy. The analyses of this paper turn this picture on its head. They help logically cement the view that, in a technologically reshaped world, it is more sensible to conceptualize entrepreneurs as restrainers of an otherwise unbridled force of machine creativity.
Notably, this reconceptualization carries important practical implications beyond theorisation purposes. First, if the bottleneck of entrepreneurship is no longer creativity but the ability to distinguish between opportunities and nonopportunities, then entrepreneurship should accordingly shift attention from creative ideation to improving judgment processes under uncertainty. Second, for policymakers, the message is equally clear: in a world where GenAI can generate new venture ideas at negligible cost, the scarce resource is judgment. Thus, rather than investing in programmes aiming to boost entrepreneurial creativity, it might be more prudent to invest in programmes that improve the ability of entrepreneurs and organizations to think more rationally in the face of the flood of “opportunities” suggested by machines. Finally, for entrepreneurs themselves, our analysis suggests that the more valuable skill in the GenAI era may not be imagination but restraint: the disciplined ability to say “no” to compelling ideas that nevertheless cannot withstand critical scrutiny.
