WELP:
pjmedia.com
Nothing at all concerning there.
Don't be worried.
Look here, I happen to know a lot about this stuff. It is EASY to emulate an ego. "Emulate" being the operative word. They've been doing this for a very long time, since the 60's. One of the first examples is Eliza, an AI psychiatrist, from back in the day when "natural language processing" was still a big deal and most of it was being done in LISP.
en.wikipedia.org
However, a REAL (or artificially real) ego is something way completely different. It is a real time superposition of two complementary reference frames, one egocentric and the other allocentric.
They're just now starting to figure out how this actually works, in a real brain. The first and perhaps best studied example we have is the visual system. First, you have stereoscopic reconstruction from the two eyes to recover depth. Then you have the extraction of 3-d boundaries to define "objects" in all their possible sizes and orientations, then you have the placement of those objects in a larger 3-d space which is bigger than just what the organism sees in front with the frontally placed eyes, it includes things like sounds that happen "in back" and "behind", and so eventually we get to where we can map an allocentric view in "external coordinates", to an egocentric view in "internal coordinates". In a human brain all of this happens inside of 200 msec.
Going the other way, the oculomotor system is equally fascinating. The supplementary eye field tells the frontal eye field "focus THERE" (on some object), and the frontal eye field will translate that into lateral, vertical, and "vergence" (depth related) vectors - those are transmitted to an area in the midbrain called superior colliculus which translates them into muscle commands for the oculomotor neurons. The side to side eye movements are called saccades, and the interesting thing about them is the vergence (depth) movements INTERRUPT the saccades in mid stream. So you have a whole separate system extracting depth and passing it downstream, and this can only occur when the source mapping is egocentric, because any other way would be computationally intractable in the time frame.
Right now we're starting to understand the mapping of sequences of events (involving objects) into "episodes" (episodes being the things that are stored in memory). It involves an ingenious form of phase coding in the prefrontal cortex and hippocampus of the brain, tied to theta rhythms which are created by the local oscillators depicted in the Kuramoto model. What happens is, if a rat learns a maze in the sequence A-B-C-D, the sequence is replayed forwards "and backwards" during consolidation - and if you look at the phase coding in a single cell, you find the forward and backward sequences overlap.
So like, a real ego is actually very much like a quantum superposition, the expectations are "not separable", there is "entanglement" - and this is precisely why we need the Renyi entropy and why information theory is an inadequate description of reality.
Mathematicians are scared of it because the special case of order=1 is the only case where mutual information is easily defined (to define other orders they'd actually have to do work which is a four letter word). However it turns out order 2 provides an exact measure of entanglement if handled correctly - and if we have all orders we can map statistical divergences directly to system dynamics. (An example for order 1 is equating the first order Kuramoto dynamic with the Kullbach-Leibler divergence). This is how learning occurs in "real" egos. There is LOGIC to it. Even rats when they learn a maze, are very logical in their explorations, very systematic.
The physics of this is very tight, you can't just put a bunch of quantum dots on a chip and call it a brain. Furthermore, proper function in adulthood absolutely depends on successful completion of the proper sequence of development. V1 "has to" wire up before V2, that kind of thing. Thirdly, new baby granule cells are born in the dentate gyrus of the hippocampus at the rate of about 15,000 a day, they take about 6 weeks to wire up and no one knows how that works.
The reason the granule cells are important is because they're at the intersection of the allocentric and egocentric reference frames. If you're interested in learning more Google "place cells hippocampus" and "grid cells hippocampus" and "time cells hippocampus".