AI-2027 forecasters move their timelines ~1.5 years earlier, predict 2027 or 2028 most likely year for AGI

A quarterly survey of AI experts, the AI Futures Forecasting Survey, has recorded a notable acceleration in predictions for the arrival of Artificial General Intelligence (AGI). The median forecast has shifted forward by approximately 1.5 years, with 2027 or 2028 now emerging as the most likely timeframe. This update, detailed in a blog post by Eli Lifland, reflects a growing consensus among forecasters that recent breakthroughs in large language models (LLMs) and AI agents are progressing faster than previously modeled.

The timeline shift is primarily driven by the observed capabilities and rapid iteration of frontier models from OpenAI, Anthropic, and Meta. Specifically, the performance leaps from GPT-4 to GPT-4o, and the reasoning improvements in Claude 3.5 Sonnet, have caused experts to recalibrate their expectations for when AI will match or exceed human performance across a wide range of economically valuable tasks. The forecast defines AGI as the point where unaided AI can perform tasks as well as a human expert.

This accelerated timeline carries significant implications for corporate strategy, national policy, and safety research. If AGI arrives by the late 2020s, the window for developing robust governance frameworks, alignment techniques, and economic adaptation plans is narrower than many institutions have been planning for. The forecast underscores the need for accelerated investment in AI safety and preparedness alongside capability development.