Roleplaying the Future
Scenario planning for high-uncertainty decisions
In this episode, Samuel Arbesman speaks with Gordon Brander , a technologist and researcher currently building Deep Future, an AI tool designed to facilitate scenario planning. Gordon has worked in design and engineering at organizations including Mozilla and Google, where he used scenario planning in practice—and he’s now aiming to democratize the framework more broadly.
Together, Samuel and Gordon explore the history and mechanics of scenario planning: what it is, how it works, and what it’s actually useful for. They dig into risk versus uncertainty, epistemic humility, and Knightian uncertainty, along with the research process behind building scenarios—and the common perils and misuses that can turn the practice into self-deception. They also discuss Gordon’s AI-powered approach with Deep Future (which Samuel is advising) and how scenario planning has reshaped Gordon’s understanding of society and the world.
Love this framwork for navigating uncertainty! The distinction between risk and Knightian uncertainty is crucial - most planning tools assume we know the probability distributions, which is rarely true for genuinely novel situations. Using AI to faciliate scenario planning makes a ton of sense given how much contextual information needs to be synthesized. My only concern is whether Deep Future can avoid the common trap of generating scenarios that just confirm existing biases.