Effectuation, causation, and machine learning in co-creating entrepreneurial opportunities

Abstract

In creating innovative entrepreneurial opportunities, entrepreneurs are characterized by limited information processing capabilities and local search routines that are immanent to humans. Machine learning (ML) offers the opportunity to overcome these limitations while reshaping the innovation process. It is indisputable that ML alone cannot realize entrepreneurial opportunities, but that close collaboration with the human entrepreneur is required. While entrepreneurs usually act according to the principles of effectuation logic in situations of high uncertainty and according to the principles of causation logic in situations of risk, it remains unclear how co-creation with ML affects the entrepreneur's decision-making behavior. By contrasting the functionalities of four different ML paradigms with the principles of the two decision logics, it is shown that supervised ML supports causation logic, while unsupervised and reinforcement ML support effectuation logic in their approach. As a fourth, semi-supervised ML is classified somewhere between effectuation and causation. However, in relation to the situational context of different types of uncertainty, ML may also prove limiting for effectuation by transitioning to causation in the medium-term.