The LLM text wall is a 21st century Plato's cave. You're staring at shadows calling it insight.
Without real 3D experience and training, the AI will always be limited. Even if you throw 2D images and videos at it, it will still be limited. It needs the 5 senses and continuos experience, which means, in the end, to have a life, in order to be really competent.
1