We Are Confusing Two Types of AGI

This article discusses the distinction between two types of Artificial General Intelligence (AGI): Soft AGI, which emulates learning generality well enough to replace knowledge workers in commercial products, and Hard AGI, which achieves true human-level learning generality from an academic computer science perspective. The author argues that most debate confusion stems from people talking about different concepts. Soft AGI is seen as an economic and societal risk, potentially replacing workers by 2026-2027, while Hard AGI remains a revolutionary but theoretical goal. The piece emphasizes the need for clear definitions in AGI discussions.