Generative AI demonstrates growing intelligence in creating realistic and diverse synthetic data, revolutionizing various industries. Different AI generation capabilities can be classified based on their level of sophistication and significance. This presentation talks about the state-of-the-art generative AI algorithms with their underlying principles, strengths, and limitations. We will discuss two AI data generation paradigms: using multiple single-task AI models or a single foundation model for multiple tasks. Throughout the talk, we will explore the potential of generative AI as a new world simulator and discuss the challenges and future directions in this field.

• Explore the growing intelligence and capabilities of generative AI in creating realistic synthetic data across industries
• Understand the state-of-the-art generative AI algorithms, their underlying principles, and practical applications in the automotive industry
• Compare AI data generation paradigms, understand generative AI as a world simulator, and discuss challenges and future directions
• Learn the advancements in 3D reconstruction, language guided world simulation, fine-tuning of foundation models with autonomous driving dataset