What's still missing, as the article mentions, is the correct relationship between objects. As in this post from a few months ago: https://astralcodexten.substack.com/i/57243633/the-tenth-vir... - it knows that Tycho Brahe was a bearded guy, and it knows what a telescope looks like, but when you ask for "Tycho Brahe looking through a telescope", the person sometimes looks at the telescope, or past it, sometimes the telescope just sits there, but the person never actually looks through it.