The Man Who Invented AGI (www.wired.com)

The story traces the origin and rise of the acronym AGI (artificial general intelligence). John McCarthy coined “artificial intelligence” at Dartmouth in 1956, but the specific label AGI — and the modern sense of it as a single-term goal — first appears in a little-known 1997 paper by Mark Gubrud. He defined “advanced artificial general intelligence” as systems that rival or surpass the human brain in complexity and speed, that can acquire, manipulate and reason with general knowledge, and that are usable wherever human intelligence would be needed. The phrase was independently reinvented and popularized in the early 2000s by researchers around Ben Goertzel and Shane Legg (with influences like Ray Kurzweil), who helped turn AGI into a conference, journal topic, and a field-shaping label. That linguistic history matters because AGI now drives major technical, commercial and policy moves: massive R&D spending at Meta, Google, Microsoft, the OpenAI–Microsoft deal, Nvidia’s valuation surge, and political concerns about geopolitical AI races and catastrophic misuse. Technically, AGI contrasts with narrow expert systems by demanding generality — learning, reasoning and flexible transfer across domains — which raises both transformative opportunities and acute governance risks (autonomous weapons, strategic instability). Gubrud’s early warning about an arms race remains salient: the term he first used encapsulates not just an engineering ambition but a set of social, economic and safety challenges the AI/ML community must confront.