Sunday, 28 april 2019 | Redacción CEU
Social networking sites are another channel of information. Nowadays, ignoring the content which is traveling on these online platforms is almost an impossible mission. Memes, stories, tweets, posts,... As our society is more immersed in this digital world, our language changes more, and so does the perception of our reality. Deepfakes are one of the last viral phenomena on the network. These hyperrealistic productions test our power of judgement. Hence, a question must be asked: Are we ready to recognize the lies when they appear on video and use artificial intelligence?
The term “deepfake” is a word game that connects its older brothers, fake news, with its underlying technology, deep learning. Deepfakes use this technique based on artificial intelligence to copy and learn the voices, tics, expressions and movements of the characters who are going to be supplanted. The result is a fake video with a very realistic appearance. They are manipulations of great quality, which are increasingly sophisticated and whose main objective is to confuse audiences. In most cases, the effect is so real that it is practically impossible to detect the deception.
One of the last contents to go viral was the video on which Jennifer Lawrence appeared talking with the face of Steve Buscemi. The effect was at least disturbing, although in this case the lie was recognizable. The goal of this video was innocent. Its creator did not even expect it would become so popular, it was just a game. Deepfakes do not always have a malicious purpose, sometimes their intention is making you laugh or drawing your attention.
The deepfake that managed to replace the identity of Barack Obama and make him say phrases that he never pronounced was more controversial. The video was created by Jordan Peele, Director of the film Get Out, in collaboration with Jonah Peretti, co-founder of BuzzFeed. It was an experiment that invited viewers to reflect: Will eventually deepfakes become a key instrument to spread lies? Most experts believe so. In the future, new technologies may play a key role in the boost of fake news.
This data is even more alarming if we take into account that researchers have come to the conclusion that fake news are 10% more likely to be shared than true facts. Largely, this situation is due to the weight that emotion has in today's society. In the post-truth era, people are more governed by passionate and visceral criteria than by the credibility of the facts themselves. This context leads to the fact that, in many cases, when users get this type of fake information, they let themselves be guided by their ideology and beliefs without questioning what they read, hear or see. In short, the recipients of the messages end up believing only what they want to believe.
Rumors, hoaxes, exaggerations, inaccurate information and lies are not new, but they are the channels through which they are conveyed, and the technological advances on which these false messages are built. While it is true that it is increasingly difficult to distinguish the lie from the truth, it is also true that the best shield is found in the critical thinking. On the other hand, this type of contents are often very controversial. In case they are true, it would be strange that any reference media did not mention them. Journalism can become a good ally in this fight against lies. Precisely, this current is reflected in the growing commitment of the media for data journalism and fact checking.
A trained eye may also be able to detect these fake videos, at least for now. Experts say that, under normal circumstances, human beings blink once every two to ten seconds. In deepfakes, the characters blink with a lower frequency. The explanation is that, in most images used to make these videos, the characters do not appear with their eyes closed. Other possible signs of manipulation are that clips are too short (the shorter they are, the more difficult to discover the deception is), the breathing is not regular, the lips are not always synchronized or the tone of voice and emphasis are not right.
Of course, these tips will be less and less useful as these techniques evolve. Likewise, lies can find new costumes to reach audiences. For example, the specialists in machine learning defend that the synthetic voices might be able to copy the human ones in few years. There are also projects like Facebook AI Research which aims to change the voice of an artist for another one in a song with just one click.
The world of digital marketing is one of the professional sectors which is experiencing a greater growth, as well as one of the areas that change most in the short term. Therefore, it is imperative to work on the training of professionals so that they may live up to the challenge. At The CEU IAM Business School, we offer a Master's Degree in Digital Marketing with a current and practical approach that responds to this need, and enables participants to design strategies that are effective in this new digital context. Thanks to its blended-learning methodology, students might combine their training with their day-to-day, and put into practice the knowledge that they have acquired from the first moment.