Nvidia CEO says "AGI has arrived" after GPT-6 Astra. Are we actually there, or are we moving the AGI goalpost again?
retained text
I saw Jensen Huang's comment about GPT-6 Astra and found the reaction interesting. He basically said: "AGI has arrived." Now, whether you agree or disagree, I think this raises a bigger question: **What actually counts as AGI?** Because the definition seems to change every few years. A few years ago, many people thought AGI meant: * passing human-level benchmarks * solving complex problems * writing software * doing scientific work But now we have models that can: * write and debug code * use tools * browse information * perform multi-step tasks * assist with research So some people argue that we crossed the line. Others say: "*These models are still pattern predictors. They don't truly understand the world.*" And honestly, I think both sides have a point. The thing I find more interesting is not whether we call it AGI. The bigger shift is that AI systems are becoming **agents**. An agent can: * decide what steps to take * use external tools * modify files * interact with other systems * complete tasks without constant human input Maybe the AGI debate is focusing too much on the label and not enough on the practical impact. My question for people building with LLMs: **What capability would make you personally say "yes, this is AGI"?** Would it be: 1. Human-level reasoning? 2. Ability to learn new skills independently? 3. Long-term autonomous operation? 4. Something else? Curious to hear opinions from people actually working with these systems. #
document fetched 2026-09-08 00:18 · canonical https://reddit.com/r/AI_Agents/comments/1w9zxuo/nvidia_ceo_says_agi_has_arrived_after_gpt6_astra