Musk Says AGI Arrives by 2026—But No One Can Define It
Get the Finance newsletter
Daily finance — markets, central banks, M&A, the prints that move money. Free.
- Elon Musk predicted "we'll hit AGI in 2026" and said he is "confident by 2030, AI will exceed the intelligence of all humans combined" in a December interview with XPRIZE Foundation chairman Peter Diamandis.
- Malo Bourgon, CEO of the Machine Intelligence Research Institute, said AGI definitions vary widely and that current models lack the autonomy most definitions require—they behave as tools and chatbots rather than agentic systems.
- Ben Goertzel, CEO of SingularityNET and one of the figures credited with coining the term, said today's large language models derive power from "having the whole internet crammed into their knowledge base" rather than generating genuinely novel insights beyond their training data.
- Goertzel argued the transition to AGI is unlikely to show up as a single clean breakpoint, comparing it to biology's fuzzy boundaries between living things and viruses or retroviruses.
- Kyle Chan, a researcher at Brookings who studies global AI policy, said AGI "is not such a big thing in China," where the AI community is instead focused on commercial applications and the physical side—robotics, drones, and autonomous systems—where China has supply chain advantages the U.S. lacks.
- The term "artificial general intelligence" was popularized in the early 2000s by researchers Ben Goertzel, Shane Legg, and Peter Voss to distinguish broad, human-level AI from narrow systems being developed in labs.
Why it matters: The disagreement has direct material stakes: Musk's 2026 prediction is driving billions in AI data center investment, while Goertzel argues current models remix training data rather than generate truly novel insights. Meanwhile, China's pivot to physical AI—robotics, drones, autonomous systems—signals a fundamentally different competitive bet that bypasses the AGI debate entirely.
Ask SkimNews
