A few weeks ago, I got a notification, Drake and The Weeknd may have just released a new track! Truth be told, these are 2 of my favorite music artists and with those 2 in a music collaboration.
So, I rushed to Spotify and listened to "Heart on my Sleeves" and truly, it did sound amazing, only to find out a day or 2 later that this just happened to be AI-generated/Deepfake music.
In interviews
Deepfake technology has been on the rise for decades and in recent years, deepfakes have become more accessible and advanced, thanks to the development of powerful and versatile generative models, such as autoencoders and generative adversarial networks. But this only makes it harder for us to distinguish between what is real and what is unreal.
In this article, we will explain what Deepfake Technology is, the rise of the technology, how it works, and its positive and negative impacts on individuals and society.
Let's explore!
Disclaimer: The section How Deep Fake Technology Works was improved by AI
What is Deepfake Technology?
DeepFakes are AI-generated videos and images that can alter or fabricate the reality of people, events, and objects. This technology is a type of artificial intelligence that can create or manipulate images, videos, and audio that look and sound realistic but are not authentic.
Deepfake technology uses machine learning algorithms to analyze and learn from real data, such as photos, videos, and voices of people, and then generate new data that resembles the original but with some changes.
Deepfake Technology Overview
Deepfake technology works by using Artificial Neural Networks (ANNs), which are computer systems that can learn from data and perform tasks that normally require human intelligence. ANNs can analyze and learn from real images, videos, and audio of people, and then generate new data that resembles the original but with some changes.
To create a deepfake, a developer usually needs two types of ANNs: a generator and a discriminator.
The generator creates fake data, such as a new face or a new voice, based on the input data, such as a source video or image. The discriminator evaluates how realistic or fake the generated data is, and gives feedback to the generator to improve its output.
This process is repeated until the generator produces convincing data that can fool the discriminator. This combination of generator and discriminator ANNs is called a
For example, to create a deepfake video of someone saying something they never said, a developer would need a source video of the person, a target audio of what they want them to say, and a GAN system.
The GAN system would use the source video to learn the facial features and expressions of the person and use the target audio to generate lip movements that match the speech. The generator would then create a new video that combines the original face with the new lip movements, while the discriminator would check how realistic or fake the new video is. The generator would keep improving its output until it passes the discriminator's test.
Types of Deepfake Technology
This can be classified based on the input and output modalities, such as images, videos, or audio.
Some of the types of deepfake technology are:
- Face Swapping: This type of deepfake technology replaces the face of one person in an image or video with the face of another person. Face-swapping can be done for various purposes, such as entertainment, parody, satire, or impersonation. A face-swapping deepfake can show a celebrity's face on a porn star's body, or a politician's face on a comedian's body.
- Face Synthesis: This type of deepfake technology creates a new face from scratch or based on some attributes, such as age, gender, or ethnicity. Face synthesis is often carried out for purposes such as art, education, or identity protection. A face synthesis deepfake can show a realistic-looking person that does not exist or a person with a different appearance than their real one. An example is Dall-E.
- Lip Sync: This type of deepfake technology modifies the lip movements of a person in a video to match a given audio input. This can be done for various purposes, such as dubbing, translation, or manipulation.
- Voice Cloning: This type of deepfake technology generates a synthetic voice that mimics the voice of a given person, often done for entertainment, communication, or fraud. AI can be used to create realistic and convincing visual content that can be used for various purposes, such as education, entertainment, art, and communication. Some of the ways AI can create visual content are:
To understand this concept better, let's consider the following scenarios where deep fake technology was used;
- A
deepfake video of former US president Barack Obama giving a speech that he never gave, created by comedian Jordan Peele. - A
deepfake video of Facebook CEO Mark Zuckerberg boasting about having billions of people's data, was created by artists Bill Posters and Daniel Howe. Deepfake videos of actor Tom Cruise doing various stunts and jokes, created by TikTok user @deeptomcruise.- A
deepfake app called Reface allows users to swap their faces with celebrities in videos and gifs.
A
The Positive Impacts of Deepfake Technology
Although Deepfake technology is widely criticized for its potential to spread misinformation, harm privacy, and damage reputations, there are still some positive impacts that are worth considering, some of which include:
-
Creation of stunning visual effects for movies, TV shows, games, and other forms of entertainment by helping filmmakers and 3D artists create realistic and immersive scenes, characters, and effects. Deepfakes can be used to revive deceased actors/stars, de-age or age actors, dub voices in different languages, or enhance facial expressions. An example is bringing back Peter Cushing in Rogue One: A Star Wars Story, or creating realistic digital avatars of celebrities, such as Tom Cruise on TikTok.
-
Protection of personal identity by concealing or altering one's identity online, such as using a different face or voice for video calls, social media posts, or online gaming. This can help protect one's privacy and security from unwanted attention or harassment.
-
Creation of realistic scenarios for professional training purposes, such as medical simulations, emergency drills, or language learning. An example is generated AI avatars that can interact with trainees and provide feedback.
-
Generation of synthetic data for scientific experiments, like testing new drugs, studying human behavior, or exploring new phenomena. It can create virtual patients that can protect the privacy of real ones.
-
Minimizing the cost and time of producing high-quality content, such as advertising, education, or entertainment. For example, it can replace expensive actors, models, or locations with cheaper alternatives.
-
In politics, deepfakes can help politicians and activists communicate their messages and causes. The technology can be used to create persuasive speeches, campaigns, or public service announcements, or to translate speeches into different languages or dialects.
-
In education, Deepfakes can help teachers and students enhance their learning experiences. It can create interactive lessons, simulations, or demonstrations, or bring historical figures or events to life.
The Negative Impacts of Deepfake Technology
Notwithstanding the positive impact of this amazing technology, there are some pronounced negative impacts that can generate ripple effects in any scenario.
These negative impacts include:
-
Undermining the trustworthiness and reliability of information sources and media platforms;
making it harder for people to distinguish between real and fake content.
-
Manipulation of public opinion or influence elections by spreading false or misleading information about political figures or events.
-
Extortion of money or favours from people by threatening to expose them with fabricated evidence of wrongdoing or compromising situations.
-
Creation of non-consensual pornography of celebrities or ordinary people; violating their privacy and dignity.
-
Creation of fake identities or accounts, such as bots, trolls, or impostors, using the likeness of real or fictitious people. It can be used to scam people by impersonating their relatives, friends, or authorities.
Final Thoughts
Deepfake technology is a powerful and versatile tool that can create realistic and convincing synthetic representations of speech and actions. This technology has had a long history with more and more efforts dedicated daily to creating better ways to make the products of the technology much more believable and almost identical.
Deepfakes have some positive implications, however, it also poses significant challenges and risks for individuals and society. Hence, more research is needed to trim down the negatives of this technology.
In our next article, we will be looking explicitly at the dangers of Deepfake Technology and the steps made to cut down on its excesses.