While technically true, it’s kind of funny how the person making the video seems surprised by the results of his own query, skipping over the top-ranked irrelevant results, sometimes just trying to rationalize what was returned.
It’s not like he’s incompetent, though. The people who “train” (procedurally throw data into) these models don’t understand them either.











I’m wondering that myself. Without digging much deeper, my guess is: