How “Conscious” are Artificial Neural Networks ?

And what does conscious mean in the first place ?

I recently stumbled upon a tweet by someone at a well known AI company that paraphrasing ( slightly ) said :

It’s possible that today’s massive neural networks are slightly conscious.

I can’t vouch either way for the author and the aforementioned massive neural net ( hence the vagueness ) but the question still remains, how “conscious” are current neural nets ? And that’s what we’ll explore here today.

Consciousness is a belief and semantic minefield, I am fascinated by what it is and how to emulate it in AI’s ( I even wrote a book , well a first part : Conscious Artificial Intelligence: Part 1. Foundations ), as I am also into Data Science,ML and ANNs ( think tensorflow ) I think I can see both ends.

For starter as I write in my book even if we made an AI that was “conscious” we would probably not recognize it as an equal ( well some wouldn’t ) and insist that there is more to consciousness than merely recreating the biological systems we possess. ( Only humans have a soul, a heart, that sort of thing ).

But if you are rational then the humbling* conclusion is that there is nothing special about…