A timeline and prediction of AI agent capability from late 2025 to 2030 by the nonprofit AI Futures Project. Likely the most detailed forecast of AI capability in the public domain, the AI Futures Project is funded by grants and donations. See
AI agent.
The predictions were formed by five researchers, feedback from more than a hundred others and more than 30 four-hour simulations called a "table-top exercise" (TTX). At the end of 2027, two possibilities are presented (SLOWDOWN or RACE). Following are the headlines of the predictions. See
AI anxiety.
AI 2027 Timeline
Mid 2025: Stumbling Agents
Late 2025: The World's Most Expensive AI
Early 2026: Coding Automation
Mid 2026: China Wakes Up
Late 2026: AI Takes Some Jobs
Jan 2027: Agent-2 Never Finishes Learning
Feb 2027: China Steals Agent-2
Mar 2027: Algorithmic Breakthroughs
Apr 2027: Alignment for Agent-3
May 2027: National Security
Jun 2027: Self-improving AI
Jul 2027: The Cheap Remote Worker
Aug 2027: The Geopolitics of Superintelligence
Sep 2027: Agent-4, the Superhuman AI Researcher
Oct 2027: Government Oversight
SLOWDOWN
Nov 2027: Tempted by Power
Dec 2027: A U.S.-China Deal?
Jan 2028: A Safer Strategy
Feb 2028: Superhuman Capabilities and Advice
Mar 2028: Election Prep
Apr 2028: Safer-4
May 2028: Superhuman AI Released
Jun 2028: AI Alignment in China
Jul 2028: The Deal
Aug 2028: Treaty Verification
Sep 2028: Who Controls the AIs?
Nov 2028: Election
--- 2029: Transformation
--- 2030: Peaceful Protests
RACE
Nov 2027: Superhuman Politicking
Dec 2027: The Agent-5 Collective
--- 2028: The AI Economy
--- 2029: The Deal
--- 2030: Takeover