DeepFakes

Obama really said that??

Over the course of the semester we have dealt with many topics, two of which are ones I was pretty interested in already and got to learn more about. These are Artificial Intelligence and Fake News, I thought that we’d also be discussing DeepFakes since it’s a combination of both. The portmanteau DeepFake is derived from 2 words (obviously :p) Deep & Fake. Deep from Deep learning which is a branch of artificial intelligence and Fake which comes from Faking something such as videos.

So DeepFakes are videos which are created using special softwares which you input multiple videos of a certain person so that the software would be able to use its artificial intelligence to learn and understand the patterns of movement of this person’s facial expressions and more specifically their lips as they talk. After that one can record a video of themselves talking which is then also inputted into the software so that the software can use it and fake the original video into making it seem like the original person is saying whatever the person with this kind of software wants him to say. The software edits the movement of his lips as well as change the voice of the creator to match the voice of the original person. This new technology has been seeing a lot of attention this last year and a half as people have used it in a lot of ways such as malicious hoaxes, or fake news, or celebrity pornography.

I believe this topic will be very useful in this course since it connects between two of the topics we talked about in this course. Moreover, I believe as DeepFakes technology and softwares progress they’ll be much more common and people should know now that it’s getting much harder to detect. Amateur-ish deepfakes are sometimes easy to spot by the naked eye by looking at the face as it sometimes looks disconnected, there is an apparent lack of blinking and sometimes weird shadows. However, the more professional ones aren’t easily recognized by the naked eye and what’s even more scary is that sometimes special softwares can not even detect that it’s a deepfake which might even create problems between countries on the global scale if there’s no way to prove it’s a fake.

Has AI gone too far?!

And here’s an article on the topic.

One thought on “DeepFakes

Leave a comment