DEEP FAKE

What is a deep fake?

Deepfake is a type of artificial intelligence used to create convincing images, audio and video hoaxes. Deepfake uses deep learning which is also a type of artificial intelligence. It uses deep learning to manipulate the image in such a way that the person won’t be able to differentiate that image from the real one. Along with images, audios can be deep faked too. The audio will be identical to the original one.

How deepfakes are created?

The person’s face is swapped with another with the help of facial recognition algorithm and VAE (variational auto-encoder – it encodes images into low dimensional representation and decodes those representations back into images). Using a VAE, the encoder will find the similarity between two images and then the images are compressed. The decoder will recover the original image from the compressed images. An image is generated by extracting the similarity from the images and from what the decoder has learnt.

What are the advantages and disadvantages of deep fake?

Deepfake can be used in movies to create a particular scene in an artistic manner. They can be used in medicine to create fake brain scans based on actual patient data to train algorithms to spot tumors in real images. Teachers can make use of deepfake to provide lessons to students in a more engaging way. Influencers or content creators can use deepfake to spread message among the audience. This will also help in increasing their audience. Forensic science can use deepfake to reconstruct the scene to understand the scene which will help in reaching evidence.

There are lots of scams because of deepfake. It becomes easier to manipulate and create fake news with the help of deepfake. These videos are very convincing which makes people believe it. People must learn to think more critically about the media they consume. For example, there was a fake video of Barack Obama insulting Donald Trump made with the help of this technology. Politicians and celebrities can be easily targeted by using deepfake. Everyone has a social media account where we share our photos frequently. People with bad intentions can get our photos easily and without permission, they might use them. So, there will be privacy problems.

How to detect deep fake videos/images?

Deepfake can be detected by noticing the facial expressions, eye movement, body posture, body movement, emotion, blurry or misalignment, unnatural coloring, lighting and poor audio. It can also be detected by comparing the video or image with the original image which we can search with the help of internet. Technology can also be used to detect deepfakes. Various organizations are working for a solution to spot a deepfake. Many big tech companies like Twitter, Adobe, Facebook, Google are already working to detect and prevent them.

There are various advantages and disadvantages to this technology. But if the disadvantages are not controlled with time, it might be destructive and people might lose hope for AI based technologies. We can make a good use of deepfake, but we need to create awareness which will help people to think about deepfake in an ethical way. Many companies are already working to detect malicious deepfake. In future, it will be of greater use in research in different sectors.

CONCLUSION

Deepfake can have positive impact in the world but at the same time there are going to be risks. It can be used to fabricate evidence or it can be used to manipulate crucial information. These risks need to be taken care of to have positive impacts of deepfake, so that deepfake can be empowering.