True, I don’t have a professional reference to directly contradict it, but that might only be because I haven’t spent any time at it, really, but just pulling things out of the news feeds that seem to have some traction on our discussions.
You could say I’m a data skeptic, I suppose. And, yes, I know there are libraries full of books detailing why evolution has made us common-sense-stupid.
Even so, I don’t think I can swallow that average, just yet. I just might not have anything more meaningful to say than, “nah.” Not because there is something essentially wrong with the whole prospect, but because I’ve seen enough of the internals of these models and systems to know that there are a hundred ways to iterate over arbitrary variables that measure things by secondary and even tertiary effects, and a lot of them are wrong, or even prejudicial.
This is an endemic problem that will remain a problem, because the obvious alternative is collecting so much data from private individuals and businesses as to deny rights of privacy. Neither sacrificing privacy, nor coercing behavior to make the models more accurate, are acceptable alternatives to the blind spots.
Even more difficult than weather, in some ways, because Nature does not go out of its own way. Socioeconomic data and AI does that all the time.