From creating an interactive image of people who have passed away to engaging customers – here are some positive uses of deepfake technologies.
Deepfake, a portmanteau of “deep learning” and “fake”, refers to hyper-realistic videos that use face swaps that leave little trace of manipulation. Deepfakes are the product of artificial intelligence (AI) applications that merge, combine, replace and superimpose images and video clips to create fake videos that appear authentic. A deepfake creator can manipulate media and replace it with a real person’s appearance (like the viral Barack Obama deepfake video from Buzzfeed).
What are deepfakes mainly used for?
One of the primary uses of deepfake technologies is pornography. AI firm Deeptrace found out that, in 2019, 96% of deepfake videos were pornographic, and 99% of those mapped faces were from deepfaking female celebrities’ faces onto porn stars’ faces. Although there are many negative effects of deepfake applications, like scamming and manipulating people, there are still many real-life applications that can bring a positive change to the world.
Positive Applications of Deepfakes
“Bringing” the deceased back to life
The most impressive use of deepfake technologies is their ability to create an interactive image and let people connect with people long after they have passed away.
For instance, the Dalí Museum in St. Petersburg, Florida, collaborated with an advertising company, Goodby, Silverstein & Partners (GS&P), to use deepfake technologies to recreate a digital reanimation of the deceased artist, Salvador Dalí. They combined archival footage from Dalí’s interviews with machine learning and trained the AI algorithm to deepfake Dalí’s face onto their actors. The final product shows an interactive deepfake video of Dalí greeting, taking selfies with visitors in the museum and reciting quotes from his interviews and letters.
GS&P Technical Director, Nathan Shipley, says, “Dalí Lives is the first time a cultural institution has used deepfakes for artistic purposes.” The interactive deepfake allows audiences to engage with Dalí’s art and the artist himself and will enable visitors to explore and experience Dalí’s artwork on a whole new level.
Snoop Dogg’s music video is another example of this “resurrection” technology. American rapper Snoop Dogg hired visual effects (VFX) production company, Corridor Digital, to use deepfake technologies to bring Tupac, an American rapper who died in a drive-by shooting when he was 25, back to life. The reasoning behind this is that Snoop Dogg wants to pay tribute to his friend and wants to “resurrect one of the greatest and most iconic music stars to have ever lived through advanced facial technology (deepfake).”
VFX artist Niko Pueringer, who worked on this project, uses clips from the MTV Video Music Awards from the 90s, where Snoop Dogg and Tupac sit side-by-side, to make the deepfaked version of Tupac lip-syncing to Snoop Dogg’s music video. Those who might not know Tupac can discover the talented actor through deepfake technologies.
Saving time and labor in the film industry
Although deepfake technologies are still in the early stages of development in the film industry, there have been some significant breakthroughs. We are now one step closer to making deepfakes more mainstream.
Most Hollywood films currently use traditional VFX, like 3D scanning and head replacements (the act of digitally replacing an actor’s head with lighting and compositing), for face-swapping. However, this process is time-consuming and often takes a lot of VFX artists to track and cut out the appropriate shots manually. But with deepfake technologies, one only needs to input videos and images they want to face-swap with into the program. AI will automatically track and replace such images in every frame. It also doesn’t need a whole team of VFX artists to do this task. As a result, deepfake technologies are the perfect solution to replace head transplants in the film industry since they can save time and labor.
Researchers at Disney have also made revolutionary discoveries concerning the resolution of deepfake technologies. They released the first photo-realistic deepfake at a megapixel resolution. Their model can now produce video at a 1024 × 1024 pixels resolution, which is a significant improvement from the current 256 × 256 pixels low resolutions. Thus, it gives deepfakes a more realistic appearance and an improved look on bigger screens.
Educating people in a more interactive way
While most online educational courses are text and graphics-based, there have been some companies and institutions who want to include deepfake technologies into their teaching materials. For instance, Udacity, an American educational organization, has started incorporating deepfake technologies into its online courses. Previously, people could only read the text slides from the course or listen to audio lectures. However, now, Udacity is investigating a new machine learning framework that automatically generates lecture videos from text-based content or audio narration. This makes the learning process more interactive than reading boring texts off a slide.
Deepfake technologies can also be used to construct artificial voices from historical figures. CereProc, a speech synthesis company based in Scotland, had analyzed 831 recordings of former U.S. President John F. Kennedy’s speeches to “build his voice” by separating the audio into 116,177 phonetic units. Since Kennedy was assassinated when he was on his way to deliver a speech in Dalla, he had never delivered his final speech to the public. But with deepfake technologies, people at CereProc were able to create a realistic audio speech derived entirely from previous audio sources. Therefore, deepfake technology can allow historical figures to tell their story/speech with their own voice to make history classes more interesting and interactive.
Engaging with viewers or customers
Deepfake technologies have also become useful in engaging customers by providing personalized recommendations and offers to meet their needs. Reuters, an international news agency, has created an AI-generated deepfake person responsible for presenting the sports news summary in the news reporting industry. People can listen to personalized news broadcasts from various news reporters. What’s more, such news reports can also be dubbed into different languages seamlessly using deepfake technologies. For instance, deepfake can entirely replace a person’s mouth movement in an English video with a French one. Compared to reading subtitles in a language that you know when the new is delivered in a foreign language, this can certainly allow for a smoother viewing experience.
In the fashion industry, many companies are now using AI to develop virtual fitting rooms where customers can scan their body and “try on” the clothes before making purchases online. In addition to that, deepfakes can also allow customers to try out the latest clothing in the virtual space—but in a different way. Data Grid, a Japanese artificial intelligence firm, allows customers to deepfake their faces onto their virtual models. With deepfake technologies, customers’ faces and bodies can now be deepfaked onto the model to see if those clothes suit them or not. This will be particularly useful for those who are shopping online.
In short, while deepfake videos are usually associated with fake news or pornography, there are still many positive uses of this technology. From bringing dead people back to life to saving time and labor, deepfake technologies have a lot of positive applications throughout a wide range of fields, including film, education and marketing. The possibilities are endless, and we are eager to see what deepfake technologies can do in the future.
- Risks Posed by Deepfake Technology and How to Combat Them
- How Do You Get Crypto Airdrops?
- The Bridge Between Man and Machine: Intersection of Ethics and AI
Header image courtesy of iStock