I think it’ll be a long time before we see something like this. It is relatively easy for computers to use deductive logic to predictively “fill in” missing features in a set of data, and to use inductive logic to generalize patterns from a set of data.
The kind of synthesis you’re referring to depends on a third kind of logic, though: the logic of abduction. With abduction, we generate (from context, experience, knowledge, and creativity) a guess that could explain what we’re looking at and then test the validity or usefulness of that guess. (If you want to think about this more, this table explains it fairly well, from this blog post). It is tough to get an algorithm to generate a creative guess, and even tougher to algorithmically test the utility or validity of that guess…
Not that it’ll never happen, but I think the human elements of this process are underrated and hard to emulate with an algorithm!
(Be careful when looking for resources on abductive logic—there’s a lot of bad explanations out there! It originated with an outcast philosopher named Charles Sanders Peirce, who was thinking about the logics of science. But images like the feature image on this post or this post—both found high up in google search results—make limited sense…)