I would like to, to borrow a recent phrase here on P-F, hit the designers of these AI algorithms until candy comes out.
Australia/Tasmania/New Zealand/PNG are the places where I most expect to find extinct animals. Though realistically the Amazon, the Congo, and other relatively inhospitable places have high probabilities as well.
But what we don't know that we don't know about animal biology is awesome. The species I worked on for my dissertation was supposedly, "nocturnal and only found in sagebrush habitat in the western US and in low abundance." - Until we started catching them well above sagebrush patches at dawn and dusk in high volume. Now it's "crepuscular with broad ecological niche in higher than anticipated abundance."
Or everything we thought we knew was wrong and everything we know now is probably wrong too...