Current Work

White Papers and Essays

Artificial Intelligence, the Complexity Trap, and National Security
June 2025

Cognitive Scaffolds for a Human Future
Interim Research Report on Agent-Based Analysis of AI’s Risks to Democratic Governance
June 2025

An Algorithm for Thought (v1.0) to Help Solve Problems of the Anthropocene
November 2022

“Special Focus Teams” to Help Solve the Problems of the Anthropocene
December 2021

 

The Crisis of Complexity and Hopes for the World of 2050

My current work takes a distinctive stance as I try to help society address the challenges of the Anthropocene. I see a dangerous “crisis of complexity” that arises from a mismatch between human cognitive capacity and the complexity of modern problems. This mismatch makes it increasingly difficult for policymakers to effectively address climate change, environmental degradation, risks posed by the rapid rise of AI, and other hyper-complex challenges of the modern world.

I’ve developed a suite of “cognitive scaffolds” designed to support human thought in this complex environment, including:

These tools are designed to support scenario analysis and policy development, and they are structured to facilitate interactions between human teams and AI-based agents that have been trained to assist in this work.

 

Putting the Algorithm for Thought to Work
Careful planning requires scenario analysis that tries to foresee how the current flow of events in the world is likely to unfold. We want a plan that can modify the flow of real-world events and lead to some better outcome. As described in my 2022 white paper, the algorithm for thought structures this as three interacting checklists: WORLD, PLAN, and OUTCOME. We want “WORLD + PLAN” to be better than “WORLD alone.”

Since August 2026, I have been using AI to build and test a working model of the algorithm for thought. The WORLD list currently has more than 300 researched features from many fields, each with sources, supporting evidence, known limits, and a confidence level. (For example, one feature notes that food for roughly half the world’s population depends on synthetic nitrogen fertilizer made by the Haber–Bosch process, for which there is no substitute at scale.) The value of such a list comes from having hundreds of features from many fields in one place, so people and AI can check how they interact with each other and with any plans for addressing the largest challenges facing humanity. This kind of shared list of explicit assumptions means that people can check the AI’s reasoning, and the AI can check theirs.

Next steps involve building out a list of hypotheses, inferences, and implications, derived by inferring likely consequences of the existing set of world features. I welcome conversations with anyone interested in testing or applying this approach.

 

The Book: Escaping the Complexity Trap
My book manuscript “Escaping the Complexity Trap: Hope for the Human Future” is currently being revised to:

  1. Focus on this new suite of tools
  2. Analyze how the rise of AI exacerbates the complexity crisis and yet also
  3. Demonstrate ways in which AI can help us work more effectively amidst the mind-boggling complexity of the modern world.

 

Credits

As I have focused on these concerns about human thought, AI, and the human future, I’ve been fortunate enough to have the advice and support of many friends and colleagues. I’ll have a more complete list of acknowledgments in the book, but – as I press to finish this manuscript  – I want to highlight the amazing figures and illustrations created by Megan Acio and the fabulous editorial help from Elizabeth Savage. 

I also am immensely grateful for help from Melissa Flagg, John Gulliver, Eric Pabo, Ken Patterson, Aditya Rajagopal, Patrick Scannell, Gurpreet Singh, Caleb Smith, Matt Thomson, Bruce Tidor, and Jeff Ubois.

THANKS! — I could not possibly do this without all your amazing help and advice!