Foundation models provide impressive abilities to generate content (text, images, sound, videos, protein structures, and more), to recognize patterns of interest and to reason at test-time, and they do so by interactive prompts in a seemingly creative manner. They hold the promise to drive innovation in many industries, changing the way we work by making tasks more efficient and providing professionals with powerful new tools.

This masterclass offers an in-depth understanding of AI foundation models like ChatGPT, DeepSeek, Llama, and GPT-5. These autoregressive, self-supervised, transformer-based models are pre-trained on extensive multimodal data and serve as the backbone of different advanced systems. They already form the basis of all state-of-the-art systems in computer vision and natural language processing across a wide range of tasks and have shown impressive few-shot learning and reasoning abilities.

Participants will explore their algorithmic foundations, emergent behavior, and ethical challenges such as hallucinations and misinformation. Gaining insights into these models enables professionals to leverage their capabilities effectively, whether in data analysis, content creation, or improving operational efficiency.

• • Pre-training at scale from language to vision and across modalities.
    • Scaling laws
    • Emergent abilities
    • Language, vision and multimodal foundation model families
  • Adaptation, -alignment and reasoning
    • Supervised and parameter-efficient fine-tuning
    • In-context learning, prompting and instruction tuning
    • Reinforcement learning from human feedback
    • Test-time scaling
  • Generating realistic multimedia content with diffusion models.
    • Diffusion model basics
    • Latent diffusion
    • Diffusion transformer and unified architectures
  • Ethical, legal and societal considerations and challenges.
    • Across the foundation model lifecycle, from pre-training, to fine-tuning and deployment
  • Use case examples across industries.

• have a better understanding of the capabilities, workings and the impact of foundation models.
• have acquired knowledge of foundation models, current theories and the core machine learning techniques to train, adapt and deploy them at scale.
• have insight into the training processes and the importance of diverse datasets.
• understand the ethical and social implications of the use of foundation models which enables you to address ethical challenges in AI implementation in your own work.
• have insight into the promising forecast of foundations models in your industry.
• can internally discuss concrete steps your organization can take to leverage the benefits of foundation models.