This AI Is The Closest Thing To AGI ( And It’s Not Chat GPT) скачать в хорошем качестве

This AI Is The Closest Thing To AGI ( And It’s Not Chat GPT)

When we talk about Artificial General Intelligence, most discussions revolve around large language models. But what if language is not the right place to look for AGI at all? In this video, we explore why some of today’s most advanced image and video generation models may actually be closer to AGI than chatbots. Using examples like Nanobanana and modern video generators, we examine a different class of AI systems that reason about the world rather than just predict text. These models do not operate purely on tokens. Instead, they build internal representations of physics, space, causality, objects, and motion, allowing them to plan scenes, understand how reality behaves, and generate outcomes that are consistent with the physical world. We discuss the idea of world models, why reasoning across multiple dimensions simultaneously matters for intelligence, and how multimodal systems that combine vision, language, and structure differ fundamentally from traditional LLMs. The video also covers what these models still lack, including embodiment, continual learning, and independent goals, and why those gaps matter when evaluating how close we really are to AGI. This is not a claim that AGI has arrived. Instead, it is a framework for thinking more clearly about what intelligence actually requires, why language alone may be insufficient, and why models that understand reality itself could be pointing in the right direction.