Abstract: The rapid advancement of artificial intelligence (AI) is reshaping how entrepreneurs generate, test, and refine new venture ideas. Concurrently, the scientific approach to entrepreneurial decision-making—which encourages entrepreneurs to think, act, and decide like scientists by building theories of value creation, formulating testable hypotheses, collecting evidence through experimentation, evaluating results rigorously, and updating theories accordingly—has attracted increasing scholarly attention. Despite the growing convergence of these two streams of research, the theoretical underpinnings of the scientific approach in AI contexts and the mechanisms through which it shapes new venture idea formation remain underexplored. To address this gap, this research proposes the concept of the “AI-augmented scientific approach to entrepreneurial decision-making” and investigates how this approach affects new venture idea formation. Drawing on the entrepreneur-as-scientist perspective, we present four interconnected sub-studies that reveal how entrepreneurs leverage this approach to iteratively refine new venture ideas toward convergence.
Study 1 introduces the concept of the “AI-augmented scientific approach to entrepreneurial decision-making” and elaborates on its theoretical underpinnings. Building on the five-step scientific decision-making framework—Theory, Hypothesis, Evidence, Evaluation, and Decision—we illustrate how AI is integrated into each step. On the one hand, AI enhances decision-making by expanding the space for theorizing, accelerating hypothesis generation, enabling market simulation, reducing cognitive biases, and improving theory-evidence consistency during theory updating. On the other hand, AI also introduces several risks, including over-reliance on data-driven theorizing, inferential disconnects between theories and hypotheses, inadequate validation due to the lack of real-world feedback, deviations from entrepreneurs’ strategic commitments, and convergence in decision-making that reduces the dispersion of theories. Overall, the effects of the AI-augmented scientific approach are highly context-dependent, varying across entrepreneur-level and task-level contexts.
Study 2 investigates the effects of the AI-augmented scientific approach on new venture idea construction during the theory-building stage. We argue that this approach enhances the innovativeness of new venture ideas by combining the systematic logic of the scientific approach with AI’s pattern recognition capabilities. However, because AI relies heavily on high-frequency patterns in training data, it tends to generate statistically probable solutions while overlooking low-probability but potentially critical insights. Consequently, it reduces idea innovativeness dispersion by producing solutions concentrated around dominant patterns. With respect to theory-market alignment, the combination of the scientific approach’s emphasis on systematic observation and AI’s sophisticated data analytics improves entrepreneurs’ structured understanding of complex markets, thereby enabling stronger alignment between value-creation theories and market needs. We further argue that entrepreneurial imaginativeness moderates these relationships.
Study 3 investigates the effects of the AI-augmented scientific approach on new venture idea testing during the theory-testing stage. We argue that this approach increases the degree of experimentation because the scientific approach emphasizes proactive evidence seeking, whereas AI enhances rigor in experimental design and reduces biases in interpreting results. Moreover, AI accelerates experimentation through automated data processing, predictive modeling, and timely feedback generation. We further argue that entrepreneurial resources—including both resource availability and resource bricolage—serve as boundary conditions that moderate these effects.
Study 4 investigates the effects of the AI-augmented scientific approach on new venture idea updating and convergence during the theory-updating stage. We argue that this approach is more likely to trigger entrepreneurial pivots because AI generates recommendations that are relatively independent of entrepreneurs’ psychological attachments, thereby enabling more objective evaluation of evidence. Moreover, this approach is more likely to induce incremental rather than radical pivots, as the scientific framework provides structured reference points for adjustment while AI enhances decision efficiency within this framework. Crucially, by influencing the iterative cycle of theory building, theory testing, and theory updating, the AI-augmented scientific approach facilitates new venture idea convergence and improves the quality of converged ideas in terms of both innovativeness and economic value.
This research makes three theoretical contributions. First, it conceptualizes the AI-augmented scientific approach to entrepreneurial decision-making by embedding AI into each step of the “Theory-Hypothesis-Evidence-Evaluation-Decision” process. In doing so, this research provides a refined theoretical understanding of the scientific approach in AI contexts and lays the foundation for future empirical investigation. Second, drawing on the entrepreneur-as-scientist perspective, this research unpacks the mechanisms through which the AI-augmented scientific approach influences the quality of converged new venture ideas. Specifically, it demonstrates how this approach shapes stage-specific variables across theory building, theory testing, and theory updating. In doing so, this research deepens our understanding of how—not merely whether—the AI-augmented scientific approach affects new venture idea formation, thereby opening the black box of how this approach works. Third, this research enriches the literature on new venture ideas by examining both innovativeness—including its dispersion—and economic value. It further conceptualizes new venture idea formation as a dynamic process in which ideas converge iteratively through cycles of theory building, testing, and updating. This dynamic view also helps clarify the boundary between new venture idea research and creativity research.

Key words: scientific approach to entrepreneurial decision-making, new venture idea formation, artificial intelligence