{"version":"1.0","provider_name":"Rainbow Dash Network","provider_url":"http:\/\/rainbowdash.net\/","type":"link","title":"adiwan (adiwan)'s status on Monday, 12-Sep-22 07:38:12 UTC","author_name":"adiwan (adiwan)","author_url":"http:\/\/rainbowdash.net\/adiwan","url":"http:\/\/rainbowdash.net\/notice\/5479784","html":"@<span class=\"vcard\"><a href=\"http:\/\/rainbowdash.net\/user\/42999\" class=\"url\" title=\"Anna Wright\"><span class=\"fn nickname mention\">oracle<\/span><\/a><\/span> I think you can make a GAN practically for everything in the realm of captchas. I think that it'll going the way I predict in order to enhance the generating models with A-B-testing (or other similar tests) by actual humans. The same as it is done now with captchas that ask humans to recognize traffic lights, cars, busses, zebra stripe crossings, bikes..."}