I guess public perception of the technology is based on how "creepy" (read: accurate) it is.

So they show erratic data on screen, preferably flattering (eg younger, happier) or just obviously wrong and record the real data on the back-end. Instant public acceptance without even writing big numbers on a red bus.

It was sinister 20 years ago (probably even before that but I'm not aware of it), and was a problem enough for the industry that they looked for way to condition the population to accept the technology over a few generations.

I don't see why it is sinister today, so they show healthy food to women and meat to men. Seems like a win-win situation.

I think this is what they did, anyway.

Why would it be sinister to have an automated eye able to instantly sort humans in dubious categories ?

Maybe in a different political context where one category is sent to death camps, another to forced labor camps, etc.

I think this becomes sinister when it makes the move to being free for the business with the service provider records & collects the data to sell to third parties.