Don’t want to drag down the enthusiasm, but we really should distinct between data analytics, actor systems, etc. and of course „artificial intelligence“.
Also, we should make a distinction between multi-layered, hierarchical information, classification and decision finding systems in general and a possible machine implementation thereof. Those systems have shown remarkable performance when operated by humans, e.g., the Dowding System [0]. (I don't think that any ML/AI systems are able to surpass this currently.)
The point is that looking at artificial intelligence like this is naive. There are some things Actor systems excel, but fail at others. There are some things NN can do pretty well, but suck at others.
Comparing humans to all applications of AI is naive and no definitive answer can be given, because it really depends. I think that this will be the status quo in the future.
This was more about comparing multi-layered systems to single actor ones and about us more or less ignoring, what humans could achieve and historically have been achieving with those, rather taking multi-layered information processing as a defining characteristic of algorithmic systems. (Actually, this isn't new, it had been deployed with massive success, and we are repurposing it with mixed results.)
[0] https://en.wikipedia.org/wiki/Dowding_system (The article contains only a quite crude and basic representation of the operations of the filter room.)