The presentation titled "AI in Engineering & Impact on Validation: Standards and Best Practices for Ensuring the Credibility of Numerical Simulations" was delivered by Vivien Clauzon and Raphaël Meunier. It focused on integrating Artificial Intelligence (AI) in engineering, particularly in high-performance computing and numerical engineering at Michelin. The presentation highlighted Michelin's three domains for sustainable growth: Tires, Around Tires, and Beyond Tires, focusing on a range of services from automobile to mobility intelligence. The speakers emphasized Michelin's approach to ensuring quality products, incorporating knowledge, product specifications, design, labs, and market considerations. They discussed the importance of fact-centric strategies, physical proofs, and achieving higher maturity levels in tire conception and material conception. The core of the presentation revolved around simulation workflows at Michelin, detailing the extensive use of user-scripting, material laws, physical models, multiphysics, and a universal file-format to bridge different software. They highlighted the challenges of errors in simulations and the importance of reproducibility and consistency across different solvers, versions, and computational environments. The speakers also addressed the role of verification, validation, and uncertainty quantification (VVUQ) in ensuring simulation credibility. They shared insights into the development and continuous integration/deployment processes, emphasizing the importance of quality assurance, sensitivity analysis, and critical difference evaluation. Furthermore, the presentation delved into the integration of AI and data science in enhancing simulation accuracy and speed. It covered case studies on augmented simulation for cost and industrialization delay reduction and the development of MIMO, a simulation-based decision-making assistant. The final part of the presentation focused on model validation in AI, stressing the need for physical validity, sensitivity analysis, error distribution, and confidence interval assessment. The speakers advocated for a multidisciplinary approach to validate models, involving product owners, data scientists, quality partners, and experts.