I'm not sure what exactly you think needs explaining. By "difficult to graps", I mean "difficult to graps that there is nothing magical going on". An AI neural network is quite dissimilar from the structures in our brain, so I wouldn't draw any conclusion from that. But I also don't see what would prevent a sufficiently large, multi-layered and recursively wired neural network to have perceptions comparable to what we experience as our consciousness.
When I instrospect my mind, everything there is, is a perception. A perception of a feeling, of an emotion, of a thought, of a recollection, of the texture of an experience ("qualia"), and of course of sensory perceptions. There is never anything that is not a perception. And a perception is just some set of information, the content of the perception.
For any given perception, it is easy to imagine that it can be represented in the brain, with its 100 billion neurons and 1000 trillion synapses, also since the carried signals and states aren't necessarily binary. It is not the case that our perceptions are more complex or more differentiatied that what could physically be represented in our brains. Our perceptory repertoire is rather finite, when you think about it. Furthermore, it is easy to imagine that our internal meta-perceptions of perceptions would arise from corresponding recursively superimposed structures and/or recursive processing in the brain.
The question is, what else, other than all those perceptions we have ("experiences"), is there in consciousness that would require explanation? I haven't been able come up with anything missing. I'd go even further and claim that there can't be anything else that we could notice or observe, since anything we observe would just be additional content of our perceptions.