Post-training and continual adaptation are reshaping how frontier language models evolve after deployment, from supervised fine-tuning and preference optimization to reinforcement learning, personalization, and self-distillation. Bringing together researchers and practitioners across alignment, safety, and systems engineering, this session examines how these methods influence model behavior, capabilities, and reliability in real-world settings. A central focus is the emergence of safety drift and anthropomorphic misalignment risks that static evaluations often fail to capture.

Advances in AI, particularly large language models, are reshaping education through personalized learning, assessment, and system-wide transformation. Bringing together academia, industry, and AI engineers, this session examines scalability, reliability, deployment challenges, and current limitations of LLMs in real-world educational settings

The growing concentration of power among dominant social media platforms is reshaping the global information environment, raising concerns around disinformation and reduced diversity of perspectives. This session highlights recent legal and technical efforts in Europe and Switzerland to address these challenges.

Cognitive scientists, AI researchers, artists, and philosophers come together to examine what “intelligence” means across disciplines and why it matters for AI design. The session explores how psychological concepts have shaped AI, questions the emphasis on abstract reasoning over embodied and social cognition, and considers whether AI reflects human or fundamentally different forms of intelligence.

AI is transforming science and engineering through generalized, multimodal foundation models that combine data-driven and physics-based approaches. Covering applications from chemistry to neuroscience, this session highlights advances such as generative AI and physical tokenization, while addressing challenges including data scarcity and model robustness

AI and machine learning are opening new possibilities for natural hazard and risk management, addressing threats such as earthquakes and landslides. This session highlights emerging methods including foundation models in Earth sciences and agentic AI for impact simulation, with a focus on digital twins and large-scale data integration.

AI is creating new opportunities for scalable impact in healthcare and medicine development, bringing together researchers, clinicians, and industry leaders to explore how innovation moves from models to practice. Through examples in precision oncology, medical imaging, and clinical decision support, the session examines data usability, validation, trust, regulatory readiness, and the integration of AI into clinical workflows and drug development through cross-sector collaboration.

Advances in generative AI are reshaping protein and small molecule design, bringing together perspectives from academia and industry to explore new approaches to discovery, optimization, and development in the life sciences.