The KDD 2026 Panel Beyond Scaling---What Will Define the Next Decade of AI? New Frontiers in Knowledge Discovery from Data

Over the past decade, the rapid scaling of data, compute, and model size has driven remarkable advances in artificial intelligence. From foundation models to generative AI, scaling has emerged as a dominant paradigm for progress. Yet, important limitations are becoming increasingly evident, including challenges in extrapolation, reliability, efficiency, evaluation, and real-world deployment. This panel will examine what comes next. If scaling alone is not sufficient, what new paradigms, system-level advances, and interdisciplinary approaches will define the next decade of AI? The discussion will explore emerging directions such as reasoning and agentic systems, as well as hybrid approaches that integrate learning with structure, prior knowledge, or explicit models of the world. It will also consider the growing importance of evaluation, trust, and deployment in real-world settings, and how these factors may influence the transition from predictive performance towards deeper forms of knowledge discovery from data, including applications in scientific discovery and other high-impact domains. Bringing together perspectives from academia, industry, and domain sciences, the panel will reflect on the evolving foundations of AI and outline key opportunities and challenges for the KDD community as it navigates this next phase of the field.