The summer of 2025 marked a definitive turning point for artificial intelligence in mathematics. Several AI models solved five out of six problems at the International Mathematical Olympiad, a feat that shocked mathematicians and demonstrated capabilities far beyond previous expectations. While Olympiad challenges involve known answers, this performance convinced many experts to experiment with AI for open-ended research. Prominent figures like Fields Medalist Terence Tao noted that AI began assisting in the discovery and proof of new results at a speed previously impossible, compressing weeks of work into days. By early 2026, the technology had evolved from a curiosity to a powerful research tool. In the February "First Proof" challenge, AI systems successfully solved more than half of ten research-level questions selected specifically because they were unlikely to appear in the models' training data. This progress signaled a shift from solving textbook problems to handling graduate-level research. Algorithms began formulating conjectures, developing proofs, and verifying results with minimal human intervention, though human oversight remained essential for validating logic and filtering errors. The collaboration between human mathematicians and AI systems yielded unexpected discoveries. In one instance, a team at Brown University used DeepMind's AlphaEvolve system to analyze permutation groups. While investigating the d-invariant, the AI uncovered a hidden structure of high-dimensional hypercubes that had eluded humans for decades. Similarly, a team at Stanford used specialized AI modules to prove a complex result regarding the embedding of spheres in flag varieties, accelerating a process that would have taken much longer manually. Researchers described the AI as a conversation partner that, despite making occasional basic errors, could offer novel strategies and verify reasoning rapidly. This acceleration has already prompted significant cultural and institutional shifts. A growing number of mathematicians are leaving academia to join tech giants like Google and OpenAI or to join specialized math-focused startups, driven by the belief that combining machine learning with mathematical precision is key to general intelligence. However, this transition raises concerns about the future of mathematical education. Experts warn that over-reliance on AI could prevent students from developing essential analytical skills, forcing universities to overhaul homework assignments and focus more on in-class verification. While the integration of AI promises exponential growth in mathematical output, experts maintain that human intuition and strategic planning remain irreplaceable for tackling the field's most profound mysteries. Tao compares current AI to a jumping robot capable of scaling small walls, whereas humans remain necessary for long-term strategic planning and navigating the highest peaks of mathematical theory. Despite the rapid pace of improvement, the consensus is that AI serves best as a catalyst that expands human capability rather than a replacement. As mathematicians navigate this new era, the discipline faces a delicate balance between leveraging AI for efficiency and preserving the artistic and creative essence that defines mathematical discovery.