The complexity and importance of advances in science and engineering have increased the need for intricate physical systems in critical operations and simulations. However, this approach is resource-intensive and costly. While AI offers a promising alternative, it remains limited in terms of generalisation, accuracy, and ethical considerations. The EU-funded TURING project will develop innovative and robust AI-driven solutions to overcome these challenges. To achieve this, it will integrate a plethora of multidisciplinary advancements from related fields to pre-train generative and multimodal foundation models capable of capturing the physics of dynamic systems sharing common properties. These models will progressively incorporate more complex physical systems to ensure robustness, ultimately becoming specialised in their respective fields.
