AI has demonstrated impressive abilities in coding, scientific research and problem-solving. Yet a new study suggests that the world’s most advanced AI systems still struggle to match elite human mathematicians when faced with original research questions. The finding comes from First Proof, a project designed to test whether AI can
