The next big leap in AI may come from systems that understand the physical world, according to Aparna Pratap, deputy general manager for infotainment at Mercedes-Benz Research & Development India (MBRDI). She said the future of AI in automotive engineering depends on systems that grasp physics, space, and context more deeply than today’s generative AI models. Pratap pointed to work by Yann LeCun on world models and Fei-Fei Li on spatial intelligence, which aim to give AI an intuitive understanding of the 3D world. She noted that current neural networks, trained on millions of data points and videos, still face uncertainty with edge cases, saying “We are always fingers crossed.” If AI moves toward world modelling and spatial intelligence, predictions could become “intuitively more intelligent” and “more natural” in physical space. Mercedes-Benz used computer vision and AI in vehicle interfaces, including gesture control as early as 2018, later deployed in the flagship S-Class in 2020. Pratap said the industry is moving beyond gesture models toward voice and new modalities. She advised young engineers to build fundamentals in core engineering, physics, and mathematics, especially linear algebra for AI/ML, and recommended a 15-minute daily reading habit, then summarising it without AI.