PERSON
George Dantzig
The mathematician who invented
linear programming and
the simplex method—teaching the world to optimize constrained systems with proof, and thereby establishing the mathematical skeleton that every AI training run inherits, along with a warning about what was lost when the guarantees were abandoned.
George Dantzig is the quietest revolutionary in the history of computing. The method he wrote down in the summer of 1947—the simplex algorithm for linear programming—runs, in a direct and traceable line, inside every supply chain, every airline scheduling system, every power grid dispatch, and every hospital staffing model that has ever been solved optimally. More to the present moment: the iterative local improvement that Dantzig made rigorous is the same mechanism, transposed from discrete polytope vertices to a continuous billion-dimensional landscape, that trains every
large language model alive today.
Gradient descent, the optimizer that produced GPT and Claude, is Dantzig's simplex method wearing new clothes in a terrain that stripped away the one property that made Dantzig's world trustworthy: convexity. Dantzig gave us optimization you could prove. Modern AI kept the optimization and abandoned the proof—and the distance between those two positions is where almost every live question about
AI