
Originally Posted by
TheFoX
Considering that images are nothing more than a sequence of pixels, how could an algorithm discriminate between a family photo of mum, dad, brother and little sister on a beach somewhere, and little sister being used for child pornography purposes?
In fact, how could an algorithm determine the difference between a home movie file you have uploaded, and a blockbuster?
It cannot. If it could, we would already have robots walking the streets with the ability to reason.
This is nothing more than Public Relations designed to scare people into thinking they have something special that they can use to detect this material. The only way to detect this material is for someone, a human, to scan it and make a reasoned judgement.
Google made the same statement about their Streetview software, stating that it automatically blanks out faces and registration plates of vehicles it photographs. It doesn't. It is done by a human. If you are observant, you will notice lots of registration plates that are legible, because someone in their haste missed them. Also, many of the faces on Streetview aren't blanked out because, again, a human missed them.
Some things cannot be left to a computer algorithm, because no algorithm can match the mind of a reasoned human, but then again, only a human could miss the obvious.
So, to summarise, this is nothing more than scare tactics to make the gullible think that they have some tool to detect transgressions on the internet. They don't. They are using our fear of getting caught against us, to frighten us into submission. If they genuinely had such a tool, they wouldn't need to announce it. They would simply trap those abusing the rules and prosecute them, extorting millions in the process.
Bookmarks