alt Hacker NewsOnline tests have very explicit grading criteria which you can be confident are applied equally. AI evaluations, as the linked video in the article points out, not only do not have explicit grading criteria but the companies promoting them can't even describe what it is.
So the meaningful difference is that unlike a test you don't know what it's looking for and you don't know if it's ranking you objectively.
Not necessarily, many online assessments I've encountered had a technical discussion session where you describe your problem solving process and design decisions. The ranking is more subjective for this component. Not to mention plenty of the assessments have more and less optimal solutions, as well as edge cases in the grading inputs that aren't in the sample inputs.