BMI is a pretty decent measure of fat percentage for the vast majority of people, it really only fails for people who have far more muscle mass than most people (body builders). It is highly correlated with health outcomes - people with a BMI over 30 (obese) have significantly higher risk of death from heart disease, stroke, diabetes, and cancer - this is true at a BMI of 30 and it only gets worse from there. 40% of americans are obese. It's not overblown - if anything, I think most people don't understand just how big of a health crisis it is.
On the margins it may not be useful, but for the "average" person, it is definitely useful. I'd suggest there is no one with a BMI of 30 (besides people with a ton of muscle) that is at a healthy weight. Its a gut check, which is what is helpful.
You know what they say, most of the volume of a high-dimensional orange is in the skin. The average person is very rare.
BMI is best for populations since it's so easy to calculate from easily and frequently collected data. For an individual, you can fairly easily get more accurate measures to predict whatever it is you want to predict.
If you're not a bodybuilder and not pregnant, it's one of the simplest useful calculations we have. You'd be amazed at how many medical decision-making tools[0] are very simple and are based on outcomes, not rigorous mathematical theory.
