What Is Deepfake? This Is A Certificate, How It Works, To How To Distinguish It From Original Content

Deepfake illustration (photo: Doc. Unsplash)

YOGYAKARTA Deepfake is one of the detrimental phenomena targeting users of information technology that continues to grow. The loss is usually a lie because it obscures the real facts. Apart from that, what is real deepfake?

In general, deepfake is a form of artificial intelligence (AI) that is used to create or create visual objects in the form of both photos and videos with the addition of audio that resembles original objects. Because it is artificial, the content produced with deepfake technology can be said to be fake.

The term deepfake itself uses the concept of deep learning carried out by AI to produce new mock products that are genuine aka fake.

The creation of deepfakes is done by combining two conflicting AI algorithms, the so-called generators and discriminators.

Generators are a tool capable of creating multimedia products. The tool will ask the discriminator to learn whether the existing multimedia products are genuine or fake. After that they will form a product called the Generative Adversarial Network (GAN) which also includes AI. The task is to find weaknesses as well as make improvements.

The discriminator will carry out the process of identifying existing visual products. If it detects the visual product is fake, it will collect important information that can be used to produce further deepfake content using GAN.

There are many ways to make deepfakes considering that many applications are available even with different qualities. But in general, the way deepfake works is to provide one's video material or targets that want to be imitated.

The maker will collect various video clips from a person or target as much as possible. The video clips collected must be related to the target. After that, AI will study the target by collecting various data, especially those related to visuals and audio.

This is where the GAN function, which is to detect, collect data, repair, and create deepfake videos whose results will resemble the original target.

In the process of creation, AI has a big role to play in determining the outcome. While humans only help collect the video clips needed.

Video deepfakes are widely used for various purposes, including the creation of hoax or fake content. An example of deepfake content was released by one of the Belgian political parties who in 2018 uploaded a video of Donald Trump while becoming president of the US.

In this content, Donald Trump deepfake is seen asking the Belgian government not to contribute to the Paris Agreement on the world climate. Whereas in the real world, Trump did not do this.

Deepfake content is also used in the film industry, such as in Fast & Furious 7. In the film, actor Paul Walker, who actually died long before the film was shown. However, with deepfake technology, Paul Walker managed to appear so as if the actor was really still alive.

How To Detect Deepfake Videos Or Not

Although very similar to the original, deepfake video content certainly has various weaknesses that can be seen. Several ways to distinguish deepfake videos are as follows.

Basically AI will create a mock object from a person. However, the resulting content still has weaknesses. You can see the skin color, hair, or other physical characteristics.

The lighting in the video is the easiest way to determine whether the content you see is deepfake or not. Usually the deeepfake content has less perfect lighting results.

How to talk someone in deepfake content can be seen easily. Usually the sound and lip movements of the object are not synchronous.

That's information related to what is deepfake. Visit VOI.ID to get other interesting information.