This week's scientific research updates highlight growing interest in agentic artificial intelligence and other AI applications transforming discovery. The National Energy Research Scientific Computing Center announced the 2026 Deep Learning for Science Summer School, which will focus on foundation models, reasoning, and agentic AI for scientific discovery, bringing together researchers and engineers starting July 20 at Berkeley Lab. Scientists also developed a new machine-learning technique called neural embedding to identify major breakthroughs in research history by analyzing approximately 55 million scientific papers and patents. This method helps locate truly disruptive innovations by measuring how far apart research papers are from their sources and their future impact. Meanwhile, researchers are using artificial intelligence in medical fields, including new AI algorithms that can classify pediatric brain tumors through liquid biopsy by analyzing DNA patterns. These developments show how AI and machine learning are becoming essential tools for accelerating scientific discovery across many different research areas.