Artificial intelligence (AI) is reshaping the way scientific discovery is conducted, from accelerating data-driven insights to enabling entirely new modes of research. As data volumes grow and experimental, observational, and simulation workflows become more complex, the integration of AI into scientific practice is no longer optional—it is transformative. This panel brings together experts from computing, domain science, and data infrastructure to discuss how AI will advance the frontiers of eScience. Panelists will explore emerging opportunities such as AI-enabled automation of experiments, large-scale model training across distributed cyberinfrastructure, and the integration of generative AI into scientific reasoning. They will also address challenges, including data quality, reproducibility, ethical and responsible AI, and the need for sustainable research ecosystems that balance human expertise with machine intelligence. By examining both current breakthroughs and long-term visions, this panel will provide the eScience community with perspectives on how AI can accelerate discovery while ensuring trust, transparency, and equity in the scientific enterprise.