I think this is distinct from "ability to perceive green, which doesn't exist". A neural network trained to distinguish green in the output of a spectroscope will perceive green without ever knowing there is something like a "neural network" or "thoughts".
Also, this probably can be detected from the outside via debugging. What cannot be detected may be the thing that distinguishes a hypothetical "philosophical zombie" from a truly conscious human, but I don't think anything like a philosophical zombie exists. Once it is physically identical to the human, it will also be thinking identically.
As a next step, you may observe humans around you and realize that the thoughts you perceive seem to be running inside the head of one of these humans (which you will call "me"). However, I don't think knowing what you look like from the outside is necessary for consciousness.
There's also an innate problem in assuming the human experience of say, "green" is consistent. What I actually see when I see the colour green only appears consistent with the physical behaviour of light. Whether any two people really see colours the same way is highly questionable.
This does not follow.
Some things that I understood, in my words:
- consciousness is probably not reducible to smaller, non-conscious parts of which it is composed. You could maybe say it is intrinsically holistic
- consciousness entails being aware of or observing qualities that hard science tells us the things don't have (green vs. length of lightwave); but "being aware of" or "observing" are so closely related to consciousness, that it may not be very informative
- consciousness can't be detected from the outside for now, and probably by the structure of the process. It is "inner" in a very peculiar sense (everything else is outer, and can't get in, except as representation)