I've noticed that the people running automated flagging systems seem to become inordinately smug to the point that they believe their false positive result over all forms of external evidence. So to them you are a criminal and that's that.
Unfortunately, the medical procedure it covered thumbnailed down (in the generated preview) to a fairly graphic photo of a woman's private parts being operated on... and that resulted in an uncontestable instaban. No humans can be reached about it, of course.
I hate automated flagging. Not only can I now not help that person on the platform in question, but I am now discouraged from even using that product further (probably not a bad thing in FB's case!)
Common argument. Not a great one, because some things are only available through Facebook, and not being there can complicate your in-person life significantly. But, since I'm not on Facebook, I have to admit that it's good enough for me.
> One needs neither of these services to live a normal life.
A step too far. Normal people have Facebook and use Google. One needs both of these services to live a normal life.
Is this because normal peoples' utility curves maximize convenience?
Also, politicians never go after companies for biased reasons and we can count on the government with more power not to abuse it.
Just because there was no AT&T style divestiture at the end, doesn't mean there were no positive externalities.
I don't know about the specifics of what is legally required vs what the Microsoft legal team decided to do to avoid further scrutiny. But the fact is that there's a lot of docs that were published in the aftermath of that ruling:
https://learn.microsoft.com/en-us/openspecs/protocols/ms-pro...
https://learn.microsoft.com/en-us/openspecs/data_portability...
The actual rendering engine was left to Windows (and to anyone who was using it, like AOL), the shell was left to IE to continue wrapping around it, and a few more things.
They weren't broken up but they were anti-competitive asswipes.
Keep in mind that the reason this happens isn't because of "faulty AI automated systems": the reason this happens is because these companies choose to save pennies by not offering proper support channels and recourse. Those saved pennies make the difference between a 2-hour outage and literally bankrupting you for no reason at all.
Wow, hello marginalization! I wonder where their training data set came from.
Which actually should be a desirable characteristic as it is an unique identifier, but it kinda makes sense with so many 'lulzhackers' out there (not that I think this is the reason - I do attribute these issues mostly to incompetence)