A new method leveraging AI techniques has been developed to explore unstable singularities in fluid dynamics equations, addressing long-standing challenges in mathematics and physics. This approach, utilizing Physics-Informed Neural Networks (PINNs), enables unprecedented accuracy in discovering new families of singularities, which could significantly advance understanding in the field.