Sagar Agarwal Member
An enthusiastic techie. Interested in Artificial Intelligence and Web Development. Love to learn more and more about technology.

AI Deepfake Technology

Quite fascinating and admittedly a bit eerie at the same time, Artificial Intelligence is now making the Mona Lisa speak!

This means that the famous “Nothing is impossible” quote is now coming true. AI is attempting to achieve the impossible as the Samsung AI Research Centre in Moscow released a study in which they use the “Deepfake” technique to bring to life Leonardo Da Vinci’s antique painting ‘Mona Lisa’. The model is documented in a paper published by Samsung AI Center, which can be read at https://arxiv.org/abs/1905.08233

Deepfakes or Living Portraits is a technique used in AI which uses convolutional neural networks to analyze and process images much like the human brain does. Traditionally, such duplicates were made using a technique called Generative Adversarial Network (GAN), in which the machine simply attempted to forge a lifelike image.

Producing deepfakes is not an easy job as human heads are geometrically complex and highly dynamic; three-dimensional models of heads have “tens of millions of parameters,” said the authors of the study. But Samsung’s new technique scans for facial “landmarks” and reduces the image to just a nose, mouth, eyes, eyebrows and a chin, to produce a sequence of entirely novel facial expressions that seem to give the images a real personality.

This technique is now also being used to create realistic videos from photos of 20th-century cultural icons such as Albert Einstein, Marilyn Monroe, and Salvador Dali!

Although this is a great advancement in AI, some people have questioned these developments as this technology could be used for nefarious purposes too. Since deepfakes are videos that realistically superimpose a person’s face and body in videos that did not originally contain them, the system can be used by a vast network of publishers and proprietors of fake news.

Thus Elon Musk, founder of Tesla and SpaceX, once expressed his concerns over AI by saying : “When there is an evil dictator, that human dies. But for AI, there will be no death – it would live forever. And then you would have an immortal dictator from which we could never escape.”