What do advances in AI mean for vehicle automation? BOX 2 AI and generative AI (GenAI) are becoming integral to vehicle automation technology. They are transforming decision-making, model training and human-machine collaboration, particularly across three key areas:3 1. End-to-end (E2E) AI models are replacing traditional rule-based systems that struggle to handle real-world driving complexity. By combining perception, prediction and planning into a single neural network, E2E AI enables faster learning and better responses across a variety of environments. While historically criticized for a lack of transparency, recent innovation is making these E2E AI models more interpretable and verifiable, resolving safety concerns and increasing industry adoption. 2. By creating synthetic data, GenAI plays a key role in training autonomous systems. Real-world data collection is costly and inexhaustive, whereas simulations using GenAI create more diverse, scalable datasets that can expose models to unusual driving scenarios. This allows autonomous systems to learn from millions of simulated miles, improving their ability to handle edge cases, such as sudden obstacles or extreme weather. However, real-world validation remains essential to ensure robustness and safety. 3. AI is strengthening human-machine collaboration through enhanced driver monitoring systems (DMS) and human- machine interfaces (HMI). DMS use AI to track driver attention, fatigue and stress, triggering alerts or intervening to prevent accidents. GenAI helps improve vehicle interfaces, enabling more intuitive voice commands and adaptive controls that minimize distractions. Moreover, by leveraging GenAI, systems become more capable of explaining their decisions, improving the user’s understanding. Regional adoption differences As Figure 4 demonstrates, personal vehicle automation will be adopted at different rates around the world. The shift is expected to be led by China followed by the United States, with Europe and Japan likely to follow a similar yet slower trajectory over the coming decade. In terms of L2+ adoption, the share of new car sales in 2035 is expected to be significantly greater in China than any other major territory. This is caused, in part, by the willingness of Chinese customers to embrace automation and domestic OEMs and suppliers’ rapid advances in automation. China is also expected to have slightly higher shares of L3 and L4 vehicles than other geographies in 2035 – the US and Europe being the only other two markets where these technologies will start to appear in privately owned passenger cars by 2035. Beyond these four key markets, the rest of the world follows a mixed trajectory, with some regions steadily adopting L2 and L2+ systems while others face economic, technological and regulatory hurdles that will slow the transition to higher ADAS/AD levels. Figure 4 also highlights the example of India, which follows a somewhat different trajectory to the other highlighted markets. In India, L0 systems are forecasted to still dominate in 2035 due to lower purchasing power and more complex road environments. India is also expected to leapfrog L1 and move more directly to L2 as this technology matures. Autonomous driving transforms cars into living spaces. ADAS/AD systems must be accessible across all regions and segments, enhancing safety and user-centric in-vehicle experiences for everyone. Gürcan Karakas, CEO, TOGG 9 Autonomous Vehicles: Timeline and Roadmap Ahead

Autonomous Vehicles 2025