DeepFake: There’s a disturbing trend of AI that could be applied to destroy your fame.
So many unimaginable things are happening in the age of advance computational that no one ever thought. Although modern computational provides numerous of oppurtunity and make researcher life easier, some maltreatment of Artificial Intelligence (AI) is being appeared as well which is extremely dangerous in the context of our society. This article will discuss some of the harmful aspects of AI and background exploration.
Artificial intelligence (AI) advances have allowed a rapid proliferation of profound fakes that are remarkably plausible, increasingly difficult to detect and relatively easy to manufacture.
Technology that did not have existence some years ago, especially in relation to strategies that use Generative Adversarial Networks (GAN), has emerged as a formidable challenge to the sound effects of the Government, Trade and Social Foundation.
What is a deepFake?
The emergence of DeepFake, or videos made using AI that can make it look like someone said or did something they never did, has raised questions about how to use such technology to expanse misleading information and destruction reputations. In recent years, high-profile personalities such as Facebook CEO Mark Zuckerberg, former President Barack Obama, and “Wonder Woman” actress Gal Godot have all featured in in-depth videos. Let’s see the video that I have trained to see how DeepFake works.
If you watch the video there, I made Emily Clark as a DeepFake and with Russian President Putin’s DeepFake was made with Donald Trump. So it is understood what kind of work can be done with these. The question that may come to your mind is how does it work? By which algorithm?
Let me clarify………………………… :p
Generalized adversarial Networks (GAN) may be used for a variety of exciting things, but what has captured the imagination of the public is the use of DeepFake by GANs, i.e. to produce videos of people talking where the face has been swapped for someone else.
How does GAN work?
By looking at the above figure, it can be clearly observed that the GAN has two important layers, Generator and Discriminator. As an example of a generator, I can give an example, suppose, you take some steps in advance to cheat with someone, it is normal that the person you are cheating with will not understand. In this case, the one who is cheating is the Generator. On the other hand, the one who is being deceived is Discriminator. Now if the one who is being deceived knows the tricks of the deceiver then no one will be able to deceive him further. In the context of a computer, The Generator takes some samples and makes a copy and when it is sent to the Discriminator, the discriminator can identify which ones are fake and which are real based on their previous learning ability.
So this is how GAN actually works and there are many advanced procedures through which these deep fakes or videos are being made and it is very difficult to detect them. It may also be difficult to use algorithms to recognise deepFake, given that those who make such videos would likely find ways to bypass such methods of detection.
Conclusion:
To conclude, in this article, I discuss the harmful aspects of AI that have grown exponentially with the technological revolution. Although there is a lot of research has been carried out in the research community so numerous of research is needed because it is very easy to copy something from a person and say something artificially through DeepFake so everyone has to do research from their own place. Any new novel solution can only fight against these difficulties.
References:
