Developments in artificial intelligence are moving at an impressive rate, so much so that it's often hard to follow. The machines have now been able to produce incredibly plausible photographs of people, but they have never been on Earth.
These realistic but non-existent faces in the photo above are the work of researchers from Nvidia. In their work on artificial intelligence. that they presented publicly last weekteam, περιγράφουν την τροποποίηση της αρχιτεκτονικής GAN (Generative Adversarial Network) to create these fictitious images. Take a better look. If you didn't know they were fake, could you tell?
In the image above you can see what progress is within four years of entering the artificial intelligence into the image. The rough black and white faces on the left are from 2014 and were published in one paper, which then launched the AI tool called Generic Adversarial Network (GAN). On the right you see a similar job that was published this month, which used the same method (GAN), but the image quality difference is clearly huge.
What is particularly interesting is that these fake faces can also be easily customized. Nvidia engineers incorporated into their work a method known as style transfer, in which the characteristics of one image are mixed with another. You may recognize the term from various image filters that have been popular in recent years in applications such as Prisma and Facebook, which can make yourselves look like an impressionist painting or a piece of cubist art.
Η application του style transfer στα πρόσωπα, επέτρεψε στους ερευνητές της Nvidia να τα προσαρμόσουν σε εντυπωσιακό βαθμό. Στο παρακάτω πλέγμα, μπορείτε να δείτε την εξέλιξη. Παίρνοντας μια εικόνα ως πηγή (source) ενός πραγματικού προσώπου (η πάνω πάνω σειρά) και αναμιγνύωντας τα με άλλα ενός διαφορετικού πραγματικού προσώπου (δεξιά στήλη) δείτε το αποτέλεσμα που παράγεται. Χαρακτηριστικά όπως το χρώμα του δέρματος και των μαλλιών κλπ αναμειγνύονται μαζί, δημιουργώντας ένα εντελώς νέο πρόσωπο.
Naturally, the ability to create realistic people through artificial intelligence raises worrying questions. These tools could be used for misinformation and propaganda and could disturb public confidence in visual elements. A fact that could damage the judiciary as well as politics. Especially in today's time when fake news has managed to fool even journalistic agencies.
As for the difficulty constructionand fidelity: It took Nvidia researchers a week to train their model on a machine with eight Tesla GPUs to create these faces. There are also indications which we can look for to find fake photos. Hair, for example, is very difficult to forge. They often look like they are painted with a brush or very blurred. However, Nvidia's work shows how fast the development of artificial intelligence in this field is, and at this rate they will not be slow to create algorithms that can prevent such mistakes.
It is obvious that there will be a battle in the next few decades about the identity of an image, between fake creatures of artificial intelligence and security researchers. And at this time, artificial intelligence has the lead.
