Re:Family fun with deepfakes
by lucideer on Friday February 02, 2018 @11:10AM
(#16291625)
They'll get more convincing. Then we'll actually get better at recognising subtle differences. Then they'll get even more convincing. Someone will develop algorithms to identify those fakes. Then they'll be further developed and improved to the point that they're impossible to distinguish at every level.
[ Parent
| on HN ]
Re:Family fun with deepfakes
by yjftsjthsd-h on Friday February 02, 2018 @11:50AM
(#16291995)
> Someone will develop algorithms to identify those fakes.
One twist: don't adversarial generative networks already work by including a "fake detector" and beating it? This cat-mouse game might last, like, all of a week.