On 23.10.19 10:11, Stefano Bagnara via mailop wrote: > PS: I REALLY don't think Microsoft is doing filtering for "unfair > competition", as I also receive Microsoft own invoices for an Office365 > plan I buy in my office365 junk folder. I simply guess SmartScreen is > somehow "out of control" (or too much a black box) and very few people > are able to check the reasons SmartScreen does something and confirm > it's doing right. I also don't want to blame microsoft for trying to > "hide things" as I've been on the antispam side, too, and I know how > much spammers can learn from few data, but I can tell Smartscreen seems > really weird from the outside and sometimes I feel there's some dice > roll behind the scene ;-)
SmartScreen sounds like it's an AI learning what's good and bad. If it is, it is probably affected by the typical machine learning problem: Nobody knows why it does it this way. It taught itself. And if you try to fix that by teaching the rights from wrongs, it usually gets wors or totally out of control. So you don't. I've seen comments about this being another weapon in the "sorry, we can't do anything about it, its automated" arsenal. I myself am pretty sure that Skynet will start just like this. An AI that decides on things and nobody knows why. It's not the "humans are bad, destroy all humans", it's more going to be like: "Kill all humans to see how this affects my decision making". Regards, Thomas Walter -- Thomas Walter Datenverarbeitungszentrale FH Münster - University of Applied Sciences - Corrensstr. 25, Raum B 112 48149 Münster Tel: +49 251 83 64 908 Fax: +49 251 83 64 910 www.fh-muenster.de/dvz/ _______________________________________________ mailop mailing list mailop@mailop.org https://chilli.nosignal.org/cgi-bin/mailman/listinfo/mailop
