In traditional industries, such as the automotive industry, incumbents must draw on big data and artificial intelligence (AI) technologies by designing AI-embedded systems integrated into their end products. While such systems are predominantly presented as paving the way for new knowledge explorative approaches, traditional industry incumbents may face challenges integrating such disruptive technology in their optimized new product development processes. Hence, this study investigates the extent to which incumbents innovate through the design of AI-embedded systems—either via explorative or exploitative strategies—by focusing on the case of the automotive industry. It employed a sequential explanatory mixed-method design and a knowledge search theoretical framework. A quantitative analysis of 46,145 patents from the top 19 traditional companies to identify AI and non-AI patents revealed that firms primarily rely on knowledge exploitation when designing and integrating AI-embedded systems, surprisingly fostering innovativeness. Complementary qualitative insights reveal that big data and AI technologies are integrated into the industrialization phase of new vehicle development, per a creative problem-solving patch. Notably, this study’s findings reveal the technical and organizational challenges limiting data-driven innovation, thereby paving a way for more technologically original innovation with big data and AI.