These times of exponential technological advancements, in both soft- and hardware, allow us to create AI’s (Artificial Intelligence) that have not been thought possible just a couple of years ago.
For example: Just look at this recent video posted by Boston Dynamics and their new AI + Robotics:
In this specific example, the little parkour exercise the robot is executing is ran from a predetermined program. But the balancing he does, how he lands so that he does not fall over, and his visuo-spatial awareness: All that is learned over hundreds of trials and errors.
And while this is all nicely interesting and very scary at the same time, Atlas is not the star of this blog.
Machines that create content
In this blog I want to focus on machines/programmes that are written and taught to make online content such as pictures, comments and more importantly: news articles and even videos.
While an AI is often very very specific in what it does (it can only do one very limited task), provided it gets fed enough data, it is (with enough iteration time), more often than not, very capable of producing a result that is similar or better than what a human would be capable of.
Now: If you give an AI enough written text of you. All your University assignments for example. Will it be able to recreate your style of writing? Your style of formatting?
What if you feed the AI a lot of pictures and videos of yourself? Will it be able to create videos and pictures that look just like you, but in reality are just artificially generated pixels?
Deep Fake – AI’s
The internet already had a lot of practice with so called “Deep Fakes”.
Deepfake is a technique for human image synthesis based on artificial intelligence. It is used to combine and superimpose existing images and videos onto source images or videos using a machine learning technique known as generative adversarial network.Schwartz, O. (2018, November 12). You thought fake news was bad? Deepfakes are where truth goes to die.
Retrieved from: URL
But while superimposing the face of Nicolas Cage onto everything that is not able to defend itself is all fun and games, Deepfakes are and will be used for malicious applications.
Because of these capabilities, deepfakes have been used to create fake celebrity pornographic videos or revenge porn.Banks, A. (2018, February 14). Deepfakes & Why the future of porn is terrifying.
Retrieved from: URL
Deepfakes can also be used to create fake news and malicious hoaxes.Christian, J. (2018. February 01). Experts fear face swapping tech could start an international showdown.
Retreived from: URL
Influencing news & politics with Deep Fakes
The following video shows a possible malicious application of deep fakes where Jordan Peele deepfakes a video of Barack Obama.
This one is particularly scary to me, where Bill Hader is imitating Tom Cruise and while doing so the face of Tom Cruise is superimposed over the face of Bill Hader by the creator of this video.
The future of online content
Deep fakes and AI in general are not going anywhere. They are here to stay and will only be getting better and more advanced.
With this in mind we need to be even more critical of the content and news that we consume. When images and audio can be faked so convincingly, one needs to be extra careful of jumping the gun on sharing, commenting or making political decisions over a newly posted video.
I guess all we can hope for is an advanced AI that can help us mere humans to spot the future deep fakes. 🙁
AI vs AI