Simons Foundation is surfacing a 15 million dollar initiative aimed at helping…

The changing relationship between mathematics and artificial intelligence raises fundamental questions about how knowledge is created, verified, and advanced. To understand why institutions are now investing in how mathematicians adapt to these shifts, it helps to know what mathematics actually does in our world, how AI systems interact with mathematical thinking, and what has historically happened when a discipline faces technological disruption. The role of mathematics in human knowledge Mathematics is the language through which we model reality. Unlike empirical fields that gather data from the world, mathematics works through proof: a statement is true if it logically follows from established axioms and previously proven theorems. This means mathematical truth is considered permanent and universal in ways that scientific discoveries are not. A theorem proven in 1850 remains proven today. Mathematics underpins physics, engineering, cryptography, statistics, economics, and computer science itself. Without advances in mathematical theory, these applied fields eventually hit ceilings. Mathematicians have historically moved slowly and carefully, building knowledge in chains where each link must be absolutely certain before the next can be added. How AI systems encounter mathematics Artificial intelligence systems, particularly large language models and neural networks, work through pattern recognition and statistical relationships rather than formal proof. They can identify patterns in mathematical text, reproduce known proofs, solve equations, and even suggest novel approaches to problems. These systems do not "understand" mathematics the way a human mathematician does; they recognize patterns in training data. However, they can process vast quantities of mathematical literature and perform symbolic manipulation at speeds impossible for humans. This creates an unusual situation: AI can appear to work mathematically without actually doing mathematics in the traditional sense. It can generate outputs that look mathematically sound but contain subtle errors, especially at frontiers where no training data exists. The historical pattern of disciplinary adaptation Fields have always changed when their tools changed. Photography forced painting to reconsider what it was for. Calculators shifted mathematics education from computation to conceptual thinking. When microscopes arrived, biology could not ignore them; instead, the discipline reorganized around what microscopes made visible. In each case, skilled practitioners had to decide what uniquely human work meant in a new technological context. Some roles vanished. Others became more valuable because they now required judgment about when and how to use the tool, not just skill at using it. Mathematics and automation Mathematical work has always included mechanical components: arithmetic, algebraic manipulation, checking proofs for errors. Computers automated much of this work over the