Google DeepMind, in collaboration with Spanish mathematician Javier Gomezano, is on the brink of solving the Navier-Stokes existence and smoothness problem, one of the seven Millennium Prize challenges with a $1 million reward. This problem, which has remained unsolved for over two decades, concerns the fundamental equations governing fluid motion. Despite their widespread practical applications—from weather forecasting to blood flow modeling—the theoretical understanding of these equations is incomplete. The challenge lies in proving whether solutions always exist and remain smooth, or if singularities, where fluid velocity becomes infinite, can occur, potentially revealing fundamental gaps in our understanding of physics.
DeepMind’s approach to this problem is groundbreaking because it leverages artificial intelligence to explore scenarios that traditional mathematics has struggled with. Specifically, the team is using AI to simulate conditions that might lead to singularities, effectively searching for counterexamples to the smoothness hypothesis. This method represents a shift from conventional problem-solving techniques, relying on AI’s ability to model complex dynamic systems and perform extensive searches through vast possibility spaces. Demis Hassabis, DeepMind’s CEO, predicts a solution could be achieved within a year, while Gomezano estimates it might take up to five years.
Beyond this specific problem, DeepMind is pioneering a new paradigm for scientific discovery that mimics the human scientific process but at superhuman speed and scale. Their method combines an “intuition machine”—a model that internalizes system dynamics—with powerful search algorithms that explore novel possibilities. This approach is exemplified by their video generator V3, which surprisingly demonstrates an advanced understanding of fluid dynamics and chaotic systems through passive observation alone, challenging previous assumptions about the need for embodied intelligence to grasp intuitive physics.
DeepMind’s broader vision extends into biology and medicine, where AI models like AlphaFold and AlphaFold 3 are revolutionizing protein folding and molecular interactions. These advances pave the way toward simulating entire cells and organisms, potentially accelerating drug discovery and understanding of diseases. Collaborations with labs such as Isomorphic Labs showcase how AI-driven search algorithms can identify novel drug compounds, transforming scientific research from a slow, manual process into a rapid, iterative partnership between human intuition and machine computation.
The video concludes with a segment called “artificial gems,” highlighting intriguing and sometimes bizarre AI projects, such as pixel art animations created with AI, robotic mushrooms playing piano, and innovative visual effects generated by V3 prompting. These examples underscore the diverse and creative applications of AI beyond pure scientific research, illustrating the technology’s expanding influence across multiple domains. Overall, DeepMind’s work signals a transformative era in both mathematics and science, driven by the synergy of human insight and artificial intelligence.
