Making GoKill Decisions for AI in NPD: The SPARK Model

Adopting artificial intelligence (AI) in new product development (NPD) offers transformative potential but poses significant challenges for engineering managers. With over 40 proven applications in NPD and 400 vendor solutions, and facing high costs to acquire and implement AI plus huge potential pay-offs, selecting the right AI applications for adoption is critical. However, low global adoption rates (23) and high AI failure rates (up to 80) highlight the difficulty in realizing AI's promised benefits. Key barriers include inadequate business cases, unreliable financial metrics, and high uncertainties, which in turn make the selection of the best AI adoption-and-deployment projects very problematic. To address these challenges, this article introduces a structured decision-making framework based on a scoring model within the RAPID gated process. The SPARK scorecard integrates proven criteria and AI-driven insights to rate AI-adoption project attractiveness, thereby facilitating gokill decisions on AI adoption-and-deployment projects for NPD. Valid scoring models, such as SPARK, counteract intuition bias, enabling more rational and objective decision-making, and have demonstrated predictive accuracy rates exceeding conventional management decision methods by almost a factor of three SPARK is unique in that it was developed by AI, with the guidance of the author.