Daniela Amodei of Anthropic cast a critical eye on the pursuit of Artificial General Intelligence (AGI). She highlights the mixed performance of current AI technologies, which succeed brilliantly in some tasks while faltering in others, challenging the notion of a one-size-fits-all human benchmark.
Google DeepMind CEO Demis Hassabis estimates AGI is five to ten years away, while Anthropic’s own CEO Dario Amodei previously suggested it could arrive as early as 2026. Elon Musk predicted his xAI company would achieve AGI by 2026—notably, after predicting it would arrive by 2025 last year. In November 2025, Claude Opus 4.5 became the first model to exceed 80% accuracy on SWE-Bench Verified, a benchmark using real GitHub issues that professional developers tackle daily.
“AGI is such a funny term,” she said. OpenAI CEO Sam Altman claims we’re “confident we know how to build AGI” and predicts powerful AI agents will join the workforce in 2025. “That’s crazy,” she said, noting the speed of advancement. “Claude still can’t do a lot of things that humans can do,” Amodei acknowledged.
Anthropic President Daniela Amodei has challenged the AI industry’s holy grail, suggesting that artificial general intelligence (AGI)—the theoretical point where machines match human-level intelligence—may already be an irrelevant concept. In a recent CNBC interview, Amodei argued that AGI no longer captures how AI systems simultaneously outperform humans in some areas while falling dramatically short in others. “Many years ago, it was kind of a useful concept to say, ‘When will artificial intelligence be as capable as a human? I think maybe the construct itself is now wrong—or maybe not wrong, but just outdated. Her comments arrive as tech giants pour tens of billions of dollars into AGI development. Amodei pointed to concrete examples where AI has already surpassed human capabilities. Anthropic’s Claude model now writes code comparable to many professional engineers inside the company itself. Yet these same systems still falter at tasks that humans handle with ease. This tension—superhuman performance in narrow areas alongside conspicuous weaknesses—undermines the idea of a single, universal general intelligence threshold.