Ever imagined The Weeknd singing a Bollywood Song?

What is it?

A type of artificial intelligence used to create compelling images, audio and video hoaxes – a combination of deep learning and fake. Deepfakes leverage powerful techniques from Deep Learning and Artificial Intelligence to manipulate or generate visual and audio content that can more easily deceive.

How does it work?

Deep learning is well known for its capability to represent complex and high-dimensional data. One variant of the deep networks with that capability is deep autoencoders, which have been widely applied for dimensionality reduction and image compression. The first attempt at Deepfake creation was FakeApp, developed by a Reddit user using an autoencoder-decoder pairing structure. The autoencoder extracts latent features of face images, and the decoder reconstructs the face images.?

To swap faces between source images and target images, there is a need for two encoder-decoder pairs where each pair is used to train on an image set, and the encoder's parameters are shared between two network pairs. By adding the adversarial loss and perceptual loss implemented in VGGFace to the encoder-decoder architecture, an improved version of Deepfakes based on the generative adversarial network, i.e., face swap-GAN, was proposed.

A conventional GAN model uses two competing AI algorithms – the generator and the discriminator. The generator, which creates the phoney multimedia content, asks the discriminator to determine whether the content is real or artificial. Together, the generator and discriminator form a generative adversarial network (GAN). Each time the discriminator accurately identifies content as being fabricated, it provides the generator with valuable information about how to improve the next deepfake.

The first step in establishing a GAN is identifying the desired output and creating a training dataset for the generator. Once the generator produces an acceptable output level, video clips can be fed to the discriminator. As the generator gets better at creating fake video clips, the discriminator gets better at spotting them. Conversely, as the discriminator gets better at spotting fake videos, the generator gets better at creating them.

Applications of Deepfakes

With the acceleration of a digital forward communication approach, audiences are resorting to a technologically advanced way of consuming information. The impetus for newer technologies such as Deepfakes is going to strengthen the way brands connect with their consumers. Deepfake technology is considered a new outset in digital branding, as it helps lower video campaign costs, create better omnichannel campaigns and provide a hyper-personalized experience for the customer.

Suppose you are someone who uses YouTube to satisfy their appetite for entertainment/ infotainment. In that case, you must have seen Hrithik Roshan in a Zomato ad claiming to have ordered from a popular restaurant in your city. This might seem a mere coincidence for the first time, but thanks to YouTube's algorithm, you are bombarded with such ads every time you click on a video promoting a different restaurant from your area.?

Sooner or later, you realize it's just a game of your device's GPS and Zomato's Deepfake ad where only the name of the restaurant and the city varies for different people. For such ads where the company makes multiple ads around a single concept, they purchase a license for an actor's identity, use previous digital recordings, insert the appropriate dialogue and create a new video. Also, if a video is needed in multiple languages, Deepfake technology can help.

Another example is from the soft drink giant Pepsi where they Deepfaked Salman Khan from the past and pitted him in front of his current self. The narrative was built on their conversation about things that changed and things that are still the same. The commercial weaved in the product relevance intelligently using Deepfake.

While Deepfake has a dark past, connecting it with scams and illegal pornography, it cannot be excluded from the evolving world of digital marketing on these grounds. The advent of advanced technologies has presented a world of possibilities. For any technology, its consequences are determined by how it is used; and this is true for Deepfake as well. The advertising industry will have to navigate new ethical conundrums and establish guidelines for what is acceptable and responsible ways to use this technology.

Sources: Various

Edited by Vidyasagar G and Govind Madhav Vyas