Andreessen declares AGI here: GPT-5.5, Claude 4.6, Gemini 3.0 match human intelligence
Marc Andreessen says AGI is already here and we missed the milestone by months.
Marc Andreessen, co-founder of Andreessen Horowitz, dropped a bombshell on The Joe Rogan Experience: AGI—artificial general intelligence on par with human ability—is already here. He claimed that frontier models like GPT-5.5, Claude 4.6, and Gemini 3.0 now answer “better than almost any expert” 99% of the time. Andreessen explained that multiple industry leaders simultaneously reached AGI about three months before the interview (early 2026). He also noted that the Turing Test, long the gold standard for AI, was thoroughly passed when ChatGPT launched in 2022, but society largely ignored the milestone.
Andreessen downplayed fears of AI replacing jobs, likening the shift to electricity and the internet—productivity growth will expand demand. He raised the “AI vampire” concern where workers may become addicted to producing more and sleep less. Looking ahead, he sees AGI powering physical AI: Tesla’s camera-only self-driving, xAI’s conversational intelligence, and Optimus humanoid robots. OpenAI CEO Sam Altman remains more cautious, calling GPT-5.5 “autistic general intelligence” in jest and noting in a blog post that AGI is now “known how to build” but shifting focus to practical applications.
- Andreessen claims GPT-5.5, Claude 4.6, and Gemini 3.0 achieve human-level AGI, with responses better than experts 99% of the time.
- He asserts the Turing Test was passed with ChatGPT in 2022, and AGI was reached simultaneously by multiple labs in early 2026.
- Andreessen predicts AGI will drive real-world robotics, citing Tesla’s self-driving, xAI’s conversation, and Optimus humanoids as key examples.
Why It Matters
If true, AGI's arrival reshapes work, productivity, and robotics—forcing professionals to rethink AI's role in expertise and automation.