I haven't changed my thinking on that, and have in fact tried to move on from there. I haven't come to firm conclusions, but an idea I am wrestling with at the moment is how we define the centre.
I believe most players, perhaps even pros, think of the centre as that space in the middle of the board. I have a sense that AI is inviting us to see it differently, and certainly not as an empty space. My feeling now is that AI "sees" an area in which weak groups exist and, because they are weak, there is an inter-relationship between them all. By their nature, what we call the corners and sides (having board edges in their support) tend not to be inherently weak and so do not have weak groups within their (fuzzy) boundaries. Groups outside this area, on the other hand, are predisposed to be weak. Rather than considering this outside area to be the empty centre (as we do), I think the bots effectively "see" instead a lerger "weak-group domain", which may occupy all or part of the geographical centre, and may even also include portions of the sides and even corners.
Within this weak-group domain a bot is able to exploit the inter-relationships in a way that is mostly beyond us, except for utterly simple cases such as ladder-breakers.
However, I see hope in the old Chinese concept of zhaoying, or call & response. Having been brought up on the local tale of Grace Darling, a fantastically brave lighthouse-keeper's daughter, I prefer to think of this concept in terms of lighthouses. A move in the centre (or as I would now more strictly say, in the weak-group domain, i.e. the perilous ocean) that shines out and offers succour to storm-lashed boats in the offing can have inestimable value. Of course it works differently in old Chinese go, because such moves have value not just in saving groups but in connecting them, which pays off handsomely in terms of avoiding group tax.
Since immersing myself in Chinese go for the Museum of Go Theory project, I have kept a weather eye open for such moves in modern go. They do occur (with less value perhaps, because of the absence of group tax, but still a significant value) but they tend not to be mentioned, and if they are, they are talked about in vague and inconsistent ways - e.g. "lends a helping hand" or "looks towards his group below." There is no concept word, no equivalent to zhaoying.
My intuition is that there should be, and that it should imply (as zhaoying does) that there is a weak-group domain. Indeed, as I write that, it occurs to me that the word for 'centre' is rare in old Chinese commentaries, and I will now pause for a moment to check....
Ta da! In my corpus of old Chinese go texts, zhaoying occurs a whopping 33 times. The word for centre appears zero times. There is no word for weak-group domain either, of course, but that is implied in the call & response meaning.
As I said, my thoughts are still in a formative stage, but already I can see value in thinking about, say, shoulder hits in terms of how they interact with a WGD rather than with just the centre. I have in mind, in particular, Go Seigen's comment that usually, when a shoulder hit is answered with a push up, the best follow up is usually not a heavy nobi (making a potentially weak group), but a jump. And if to comes to that, thinking of a jump as a barrier, in the Chinese way, rather than as a jump, has value, too. But, of course, learning to identify a WGD allows you also to make repairs, or prophylactic moves. Such moves feature heavily in AI and pro play, and often not at all in amateur play. When amateurs do play them, it seems to me it's often a case of monkey see, monkey do rather than with true understanding (as with josekis). And when there is some understanding, it is limited to
one weak group rather than the weak-group
domain. There is no call & response - just a response.