Deepfake: Machine learning Threat or Strength
What is Deepfake?
It is a form of Artificial intelligence and it is combined term of “Deep learning” and “fake”.
Deepfakes are fake or manipulated videos or audio recordings that just look and sound like the real one. It uses facial mapping technology and Artificial intelligence (AI) that swaps the face of a person in a video into the face of another person, for this large set of data samples are given as an input to learn to imitate a person’s facial expression, voice. Deepfake came into focus in 2017 when a Reddit user posted raunchy videos of celebrities.
How Does It work?
Deepfake content is made by using two advanced AI algorithms in which one is called the Generator and the other one is Discriminator. Combinedly, the Generator and Discriminator form GAN(Generative Adversarial Network).
The Discriminator identifies the content as being manipulated, it provides the generator with precise information for how to improve the next Deepfake, using this self-learning method every time the Deepfake content gets more accurate and made looks like real.
How to detect Deepfake Videos?
Deepfake detection is a quite difficult problem. Clumsy deepfakes can be detected by the open eyes. Deepfakes generated using GAN are getting better progressively for which we have to rely on Digital forensics.
Is there any benefit of Deepfake?
This AI-based technology has disadvantages affecting different groups of our society. Apart from creating fake news and propaganda.
Though it is harmful to society it also has few advantages like creating extraordinary attention among the online audience making the web page popular on the search engine.
As every technology has is its own pros and cons, so to spread social awareness in society it is the best solution to prevent any harm from such technology.
