the secret to solving the royal game

Alpha zero has blown all our minds. As chessplayers- it surely should. Many of us must be wondering- with such a quantum leap in playing ability- when will chess become solved? I am here to provide a hypothetical next-step in such a theoretical possibility. All of this is speculation, and like the way alpha zero plays chess, mainly intuition. Many of you fellow players may think such speculation is redundant, uneducated, or even impossible. I welcome all such arguments as this debate is an important one. If you think the debate itself is redundant: go to another forum.

People are describing the play style of A0 as "alien". Is it? I, and many others, noted its much more agressive  (and successful) style as more human (not to mention its search algorithm methods i.e. 80,000 positions per second vs tens of millions for modern engines [only the good continuations are to be evaluated]; this is much more human-like). I am not trying to debate that we can play like alpha- we can't. Try and debate that here and you will be trolled for your ignorance. I think we all agree however that as alpha zero is another step toward chess perfection- it is still far from perfect. Its 24 losses of 1200 games starting from "sound" theoretical, popular human openings that have developed over time shows that it is indeed beatable and has not discovered perfect chess theory. My question is- how ever many dozens or hundreds of years from now- what will?

In my opinion the next step is engineering; this time, from the AI itself. If we develop a program that can- itself- program other rules based engines- or rules based AI (for those of you who think terminator could soon follow I included "rules based")- such a machine could program a machine without human limits of interpretation. I'm not talking about an AI that could teach itself but rather one that could build a mistake-proof program that could extrapolate data from the highest level games (exclusively non-human unless we're talking about extensively-analysed immortal games) and grow from that. Imagine a newer stockfish or komodo built off of alpha zero's intuition that could search 80,000,000+ positions per second, but on fewer variations (ie higher depth values) and could create brand new intuitive and ingenious concepts. Is this the next step for AI and chess progression? What do you think?