Be Right Back, Anthony Bourdain
When we allow Artificial Intelligence to create deep fakes and speak for the dead, it becomes an ethical question.
We mourn the death of celebrities, as most of us were saddened when Anthony Bourdain died on July 11, 2018.
The new Anthony Bourdain documentary deepfakes his voice.
This year, a documentary on his life is receiving mixed reviews. However, it also leads us to ask this question, Is it ethical for the documentary to deepfake his voice?
When Helen Rosner of The New Yorker asked Oscar-winning filmmaker Morgan Neville how he found an audio recording of Bourdain reading his own e-mail, the filmmaker replied;
“But there were three quotes there I wanted his voice for that there were no recordings of,” Neville explained.
So I got in touch with a software company, gave it about a dozen hours of recordings, and,
“I created an A.I. model of his voice.” — Morgan Neville.
What is deepfake?
If you watched the Queen’s “other Christmas message” or the viral Tom Cruise video on TikTok where he played a magic trick and was licking a lollipop, you have seen a deepfake — both are samples of deepfake videos clearly for entertainment.
Although if you watched Channel 4’s alternative Queen’s Christmas message, it is clear that at the end of the video, people would understand that it is a deepfake video with only a deepfake actor playing the Queen.
“ It will take more from us to trust what we see, hear and watch.”
A deepfake is a new breed of video that became popular with certain online communities over the past few years. These fakes use AI technology to transplant one person’s face onto another person’s body. A piece of software takes as many images, and videos of the target’s face as then uses that to create a special map. The creator can then apply this map to any already existing piece of footage. — Excerpt, What Are Deepfakes And Why Are They Dangerous?
Is it just about videos?
Deepfakes are not only about videos. For example, in The New York Times article — Designed to Deceive: Do These People Look Real to You?
Generated.Photos, one of the companies profiled, is where you buy a “unique, worry-free” fake person for $2.99, or 1,000 people for $1,000.
While the website clearly states in their user agreement,
Namely, you can not use our photos to conduct any sort of illegal activity, such as defamation, impersonation, or fraud.
Who can stop people from using these “fake people” in their nefarious activities?
One example clearly shows that the purpose has already been served before the truth was exposed when a fake journalist was published in different international publications.
The identity of which was later revealed as a deepfake. The journalist is nothing but an AI-created persona.
How are they made?
We have seen it before in movies. For example, I remember watching Forrest Gump, where the character is superimposed at different times in history.
Dead celebrities appear in TV commercials or movies, all using Computer-generated imagery or CGI, which blends with how people are creating deepfakes today.
But new technology has democratized the process, that it now takes less skill and less time to produce deepfake projects.
This is possible with a new type of artificial intelligence called generative adversarial network or (GAN).
A generative adversarial network (GAN) has two parts:
1. The generator learns to generate plausible data. The generated instances become negative training examples for the discriminator.
2. The discriminator learns to distinguish the generator’s fake data from real data. The discriminator penalizes the generator for producing implausible results.
When training begins, the generator produces obviously fake data, and the discriminator quickly learns to tell that it’s fake.
Manipulating video is nothing new — Face Swap videos.
The process calls for an encoder; through an A.I. algorithm, it learns the similarities between two faces and compresses the images. A decoder then recovers the image. By feeding the encoded images to the “wrong decoder, the decoder reconstructs the image and creates face-swapped images that can then create deepfake videos.
Are deepfakes always malicious?
They are mostly for fun and entertainment. And for the geeks out there, it can mean “Family fun with deepfakes.”
From the movies to porn where deepfake become as ubiquitous as celebrity home sex tapes, Reddit communities are banned for using deepfakes after the social media company recognized that deepfake porn is involuntary porn.
From entertainment, deepfakes found themselves a political tool to amplify fake news, conspiracy theories, and revisionist history.
And this is when Deepfakes can become dangerous to our society when it is used to undermine democracy and truth.
So the questions that follow are:
What happens if we can no longer trust our eyes or our ears?
What if we can dismiss real events as fake?
What’s next?
What about fake audio?
Deepfake audio is harder than photos and video, as it requires using someone’s voice in proximity to the voice of the person you are recreating a deepfake audio recording.
How do you spot a deepfake?
See how deepfakes are different — . Computers, not humans, do the hard work as what Hany Farid calls AI synthesized fakes.
But, instead of humans doing countless hours laboring on creating deepfakes, A.I., through deep neural networks, computers synthesize the images to create a deepfake photo, video, and even audio.
Roadrunner by Morgan Neville
If the reporter didn’t ask the filmmaker, he wouldn’t have disclosed that the documentary deepfakes Bourdain’s voice. So it brings us back to the question, is there an ethical use of A.I. and deepfakes?
Henry Ajder, a deepfakes researcher, asks a similar question in an interview with The New York Times.
“In what cases do we need consent of the deceased to resurrect them?”
In the documentary, Roadrunner, Bourdain or a deepfake voice of Bourdain, uttered the phrase;
“. . . and my life is sh#t now. You are successful, and I am successful, and I’m wondering: Are you happy?
The Verge points out that you can actually hear the deepfake voice in the trailer of Roadrunner on this YouTube video, at around the 1:30 mark.
Anthony Bourdain didn’t say that. Instead, he wrote those words to his friend, David Choe, who is in the best position to decide how Bourdain would have said those words if spoken.
And while, the message even by reading Bourdain’s words, could mean he was sad and asking for help, it can also mean that he, as most of the people who knew him says, that Bourdain is searching.
What a deepfake audio can’t capture is the emotions. The same word said differently, or which syllable is stressed takes a different meaning, homographs as we call them.
Who is making deepfakes?
While filmmakers come to mind making deepfakes for entertainment, their artistic freedom doesn’t allow them to recreate imagery of their own perception of what a dead person said.
A.I. Ethics and Why it is important.
When pressed, the filmmaker said,
“We can have a documentary-ethics panel about it later”
Full disclosure — In the already confusing world we live in, where truth becomes a scarcity. The world doesn’t only belong to the living but also the memories of the dead.
All it takes is to make full disclosure when using deep fakes if there is no malicious intent.
And when did our written words become our speech?
The non-disclosure by the filmmaker betrays the very man the documentary is trying to understand.
A.I. — It is what we do with it that makes it wrong or dangerous
Deepfakes can also become a tool to spread a message of hope, or a message that helps bring in change or an urgent plea to end gun violence, as in the case of this video showing Parkland victim Joaquin Oliver comes back to life in a heartbreaking plea to voters.
Conclusion
A.I., like any life-changing technology, will not destroy humankind or lead to a dystopian society. Instead, it is us who will use technology to destroy and harm ourselves and society at large.
Technology is moving at a pace that needs regulation.
Be Right Back, Anthony Bourdain
In an episode well remembered by Black Mirror fans, the episode's title is Be Right Back — a story of grief and how artificial intelligence is used to bring back someone from the dead.
In this case, the loving husband, and yet, in the end, the grief never ends, because the deep fake husband gave her the uncanny valley feeling, that when it feels so real, and yet it brings your feelings from empathy to disdain.
The same goes for the documentary Roadrunner. It was all good until it gives you the uncanny valley feeling.
It started to be like food deconstructed that doesn’t appeal to the palate, and for the people who loved Anthony Bourdain, if there was a lesson they could have learned from him, it is to keep parts unknown.
