AI Came for Math. Now It’s Chasing the Rest of Science (www.wired.com)

At a recent MIT talk, Neil Thompson shared groundbreaking insights on how AI is poised to revolutionize multiple scientific fields beyond mathematics, including statistics, economics, and high-energy physics. His study showed that between mid-2024 and mid-2026, the most advanced AI models improved their problem-solving capabilities in these domains from 2% to 35%. This rapid progress indicates that AI is not just automating existing tasks but is set to transform theoretical inquiry and problem-solving across science. The significance of these advancements stretches beyond mathematical solutions, sparking debates over the role of human experts in interpreting and validating AI-generated results. As AI begins to tackle complex theories autonomously, the scientific community may face both opportunities and challenges. While AI could lead to faster theoretical breakthroughs, it raises concerns about the accuracy of solutions and the potential for AI to make mistakes in complex scenarios. As some researchers advocate for increased AI involvement in experimental sciences, the need for human oversight remains crucial to ensure that AI solutions are sound and meaningful.