Deep Future is an AI initiative focused on strategizing for a range of possible futures using a scenario generation engine. The project draws on Cold War-era research by RAND, which applied game and systems theory to military strategy. Instead of making precise predictions, it explores multiple scenarios to better navigate extreme uncertainty, a concept differentiated from risk. Risk involves known outcomes with assignable probabilities, while uncertainty encompasses unknown outcomes, complicating decision-making. The article classifies uncertainty into four categories: clear, complicated, complex, and chaotic environments. Clear environments have predictable cause-and-effect relationships, making them manageable through decision theory. Complicated environments require expert analysis to understand known unknowns, allowing for educated guesses. Complex environments introduce unknown unknowns and nonlinear behaviors, where strategies based on linear thinking can lead to unintended consequences, as illustrated by the disastrous campaign against sparrows during The Great Leap Forward in China. Chaotic environments lack structure, demanding rapid responses without clear strategies. Different environments necessitate different approaches. In clear situations, optimization and decision theory thrive. Complicated settings benefit from superforecasting and expert analysis. Complex environments require adaptability and awareness of feedback loops, while chaotic contexts call for quick reflexes and survival instincts. Understanding these distinctions helps strategists apply the right tools in varying circumstances, enhancing their chances of success amid uncertainty.