About Us

The AI 2040: Plan A scenario is the third major release from the AI Futures Project, alongside AI 2027 and the AI Futures Model. We're a research nonprofit working to forecast the future of advanced AI and (now) offer recommendations for navigating it. We created this website in collaboration with Lightcone Infrastructure.

Contributions

Thomas Larsen started the project and had final say on its content. He helped shape many of the core ideas and did a large fraction of the writing, including many of the supplements.

Romeo Dean led the verification research and the compute and economics modeling, and wrote the corresponding supplements along with helping to shape many related ideas.

Daniel Kokotajlo came up with many of the core ideas and did a large fraction of the writing for the main scenario and perspectives.

Ryan Greenblatt came up with many of the core ideas behind Plan A and gave extensive feedback.

Eli Lifland did project management, helped edit the main scenario, and wrote several supplements.

Brendan Halstead led the covert project modeling and wrote the corresponding supplement, as well as the takeoff supplement.

Lauren Mangla helped with project management and a wide variety of aspects of the project, including the website, user interviews, and leading the comms.

Raymond Arnold, Oliver Habryka, and Joanna Bregan of Lightcone Infrastructure built and designed the website and provided thoughtful feedback on the content, helping make the text and the communication of ideas much better overall—the foreword, opening of the scenario and deal section especially benefited from Raymond's improvements.

Scott Alexander rewrote various sections of the plan and provided extensive feedback.

Nicole Sanna helped with the comms, project management, and a range of operations tasks.

Miles Kodama rewrote the insider's perspective, triaged feedback, and researched legal authorities and other near-term helpful actions.

Josh Clymer helped with some of the early work around verification and dark compute.

Tom Houlden helped draft an early version of our economics model, and Tom Cummingham also provided helpful high level input on the economics modelling early on.

Mia Taylor and Fin Moorhouse of Forethought wrote the space governance supplement.

Vy Tran and Dillon Nguyen reviewed various supplements and provided operational support.

Dave Kasten and Eli Tyre for helping with user interviews, comms and other supporting tasks.

Nick Marsh helped research about near-term domestic policy ideas. Jeffrey Wang, Craig Jolley, and Liam Patell helped research likely government responses.

We used AI to help with many parts of the creation of Plan A, including for: programming, brainstorming pieces of text, and giving and summarizing feedback. Human authors wrote nearly all text in the final publication, and reviewed/edited the small amount of AI-written text.

Finally, our warm thanks to everyone who gave helpful feedback in their personal capacity: Aalok Mehta, Aaron Scher, Ajeya Cotra, Alex Kastner, Alex Mallen, Alexandra Bates, Andrey Fradkin, Andy Masley, Aniket Chakravorty, Anjay Friedman, Anson Ho, Ben Chang, Ben Goldhaber, Ben Harack, Brendan Steinhauser, Daan Juijn, Damon Binder, Daniel Filan, Daniel Reuter, David Rein, David Udell, Douw Marx, Edouard Harris, Eric Gan, Finn Hambly, Fynn Heide, Grant Demaree, Halfdan Holm, Helen Toner, Holden Karnofsky, Ida Caspary, Jaime Sevilla, James Sanders, Jason Hausenloy, Jeremy Gillen, Joe O'Brien, Jonathan Erhardt, Jonathan Mann, Joshua Jacobs, Juan Felipe Ceron Uribe, Julie Lifland, Jun Shern Chan, Kristian Rönn, Leo Gao, Linch Zhang, Luke Dawes, Luke Marks, Malo Bourgon, Max Harms, Max Nadeau, Michael Trazzi, Nate Soares, Nick Gabrieli, Nicky Case, Nikola Jurkovic, Noah Ringler, Oscar Delaney, Owen Cotton-Barratt, Philip Trammell, Richard Ngo, Roy Rinberg, Saif Khan, Sasha Krecinic, Seth Lifland, Simon Steshin, Steven Veld, Sydney Von Arx, Tao Lin, the team at UK AISI, Tim Hua, Timothy Kokotajlo, Tom Davidson, Tom Milton, Trevor Levin, Varun Gandhi, Will MacAskill, Yafah Edelman, and Yannick Mühlhäuser.

This project was a deeply collaborative effort, and we're grateful to everyone who helped bring it to life.

Get in touch

You can reach us at info@aifutures.org. We look forward to hearing from you.

If you’re interested in more of our work, you can subscribe to our Substack here. You can also learn more about us and the team on our website.

If you’re interested in working with us, you can fill out an expression of interest here. If you’re interested in helping make AI go well for humanity more broadly, check out this post on how to get involved. And if you’re interested in the kind of technical verification work featured in this plan, including tools that could help enable a trustless deal with China, here are some resources on how to get involved in the AI verification field.

We accept donations online or through major DAF providers.