With input from Vitalik Buterin, Adam Marblestone, Christine Peterson, Andrew Trask, Anthony Aguirre, and others, we’ve just published AI Pathways, a project from Foresight Institute’s Existential Hope program. It explores two detailed, plausible futures for AI by 2035:
The Tool AI Pathway: a world shaped by powerful but controllable systems with limited agency, designed to extend human capabilities rather than act autonomously.
The d/acc Pathway: a decentralized, democratic, and defensive acceleration of progress across science, governance, and infrastructure.
Both scenarios are mapped with timelines, tensions, and enabling conditions. They’re meant as concrete starting points for critique, iteration, and stress-testing, not predictions or endorsements.
Two new AI futures mapped out: Tool AI and d/acc (linkpost to EA Forum post)
Link post
With input from Vitalik Buterin, Adam Marblestone, Christine Peterson, Andrew Trask, Anthony Aguirre, and others, we’ve just published AI Pathways, a project from Foresight Institute’s Existential Hope program. It explores two detailed, plausible futures for AI by 2035:
The Tool AI Pathway: a world shaped by powerful but controllable systems with limited agency, designed to extend human capabilities rather than act autonomously.
The d/acc Pathway: a decentralized, democratic, and defensive acceleration of progress across science, governance, and infrastructure.
Both scenarios are mapped with timelines, tensions, and enabling conditions. They’re meant as concrete starting points for critique, iteration, and stress-testing, not predictions or endorsements.
We also worked with Metaculus to publish forecasting questions tied to key milestones, with a $5K commenting prize pool.
👉 Full write-up on the EA Forum: https://forum.effectivealtruism.org/posts/GaFayvcDRCkvFkWri/foresight-institute-launches-two-possible-future-scenarios
👉 Scenarios: https://ai-pathways.existentialhope.com/
👉 Metaculus tournament: https://www.metaculus.com/tournament/foresight-ai-pathways/