Finally, a solution to scam email!

Sorry Mike, I don't want to attack you personally, but for instance the example of Google's image recognition algorithm shows that you don't have a clue of what AI or neutral net means. You let yourself be scared by something you don't understand, enhanced by the media reporting. Of course this is a natural human response, so I don't blame you for that.
I use a deep learning algorithm to label my own photos at home. It makes the same mistakes as you mentioned, mostly due to inadequate training of the network. Badly it is nothing more than an enhanced and more powerfull version of pattern recognition algorithms we have been using for the last few decennia, based on HMM models.

Marcel, absolutely NO problem whatsoever. If you stick to English there is nothing you could call me that I have not heard before. And I'm impossible to insult so say whatever you want.
You are correct in that A.I. and Neural Nets are not my field. In my day we were still searching for intelligence among humans.
As I posted, only half jokenly, I'm a Ludite. NEW is not necessarily improved and or better.
If we as humans create aware A.I. why would it be any better than we already are and Man's inhumanity to Man is as old as the race itself. Then enter the unintended consequences..."Gee Dundee let's import rabbits to the Outback-great hunting." "These Asian walking catfish make great outdoor pond fish"
And it does not have to be human level intelligence. Cockroaches are very successful critters. What about a LAW with the intelligence of a cockroach?
Autonomous A.I.s are very scary and I'm not alone in that assessment