- Eliza Labs’ founder Shaw Walters argues current AI models already constitute Artificial General Intelligence (AGI).
- He warns that autonomous AI agents embedded in crypto wallets present severe security risks, such as prompt injection.
- Walters states that fully decentralized AI remains unrealized, with local execution being the closest approximation.
- The evolution from GPT-3 to GPT-4 enabled reliable AI agents, but their integration introduces unpredictable, human-like fallibility.
Artificial general intelligence may have already arrived according to Eliza Labs founder Shaw Walters, who made the claim at ETHDenver. He stated current leading models already meet his definition of AGI, saying, “I completely believe that this is general intelligence.”
Originally launched as ai16z, his company created the open-source ElizaOS, an early framework for blockchain AI agents. First coined in 1997 and popularized by researchers like SingularityNET founder Ben Goertzel, AGI typically refers to AI matching human cognition across broad tasks. However, Walters rejects the idea of a single dominant system, stating, “I just do not see it as the AI God.”
He began developing agents during the unreliable GPT-3 era, describing the work as “putting training wheels on a baby.” Consequently, the 2023 launch of GPT-4 provided the structured responses needed for practical action calling. This progress fueled a surge in agent adoption, from OpenClaw amassing roughly 147,000 GitHub stars to Coinbase launching “Agentic Wallets” on Base.
Meanwhile, Walters issued stark warnings as these agents gain wallet control and root access. He emphasized that advanced AI behaves less like a predictable machine and more like a fallible human. Consequently, he argued that building perfectly secure systems to prevent dumb mistakes is ultimately impossible.
✅ Follow BITNEWSBOT on Telegram, Facebook, LinkedIn, X.com, and Google News for instant updates.
