Last year, the Beatles released the song “Now and Then”. The song and video sequence in the voice of John Lennon rekindled old memories among a generation of music lovers. It quickly garnered attention and accumulated millions of views. The Beatles fans poured their emotions into the YouTube comments as if the Beatles made history again. There was only one anomaly though in this stimulating experience.
The song “Now and Then” was written and sung by John Lennon but wasn’t recorded because of his murder in 1980. Paul McCartney who received the demo tape from Lennon’s widow decades later tried to record it. But the surviving Beatles never completed it.
How did the “last song” of the Beatles get released more than 50 years after the legendary band broke up?
How did the voice of John Lennon rendering “Now and Then” soothe the Beatles fans 43 years after his death?
Most shockingly, how come the dead members of the Beatles appear in the video? Is it disguise?
It is said that “Seeing is believing”. No more.
Deepfakes are nothing but high-end computers using AI to clone an image or video. To clone a voice, data is fed to train the AI models to generate authentic voice which can be used to make it speak anything you want.
In this case, John Lennon’s voice recordings from his heydays were used to generate voice clones of the lyrics he had written before being murdered.
Similarly, the videos were also cloned adjusting for the technology and age difference of the Beatles. Only the surviving Beatles McCartney and Ringo Starr are real in the latest release. With Artificial Intelligence (AI), the barriers of space-time perceived as immutable for time immemorial seem surpassed, at least in the digital world for now. As one Beatle fan aptly commented, “We can now officially tell our grandchildren that we lived in the times when The Beatles were making music.”
Deepfakes powered by AI is making seemingly impossible things possible and blurring the distinction between fake and real. As the ability of AI keeps increasing, deepfakes pose a tremendous opportunity for mankind. While AI was used for a constructive purpose in the case of Beatles video, such technology also poses the risk of deepfake voices, identity theft, and impersonation. Apart from being disorienting to see the two departed Beatles joking around and smiling with their mates in the video, such technology poses risks to users when in the wrong hands. AI not only creates convincing fakes that can rival the real thing, it can tamper with illusions of birth and death.
Big AI labs of today are equivalent to nuclear labs of earlier eras which discovered nuclear power as well as the bomb. We have entered a similar era with AI where the constructive and destructive use of it can be catastrophic, while we focus on the productivity increase and other benefits of AI we need to control the negative impacts it can deliver. Fortunately, in the case of nuclear fission, the horrors of Hiroshima and Nagasaki prevented another destructive use while everyone gained its positive benefits:
The positive use case of AI taps into its ability to create at a fast pace. A writer aided by AI can type faster and put words to his thoughts quickly without grammatical errors.
A designer can obtain a first-cut design merely by entering a command in the text. Essentially, writers and designers start strong with AI. Customer service is enhanced and costs lowered when an AI Chatbot can simply respond to a human client and offer predictable results without getting tired or feeling mood swings.
The negative use cases of credible-looking AI-generated photographs or videos is a danger of it being produced in court and ending up framing innocent people. An extreme case could be an AI-powered armed robotic soldier designed for war going rogue and killing civilian people indiscriminately. If data fed into an AI model is biased, it can have unforeseen consequences including accentuating societal fault lines like racism, anti-Semitism, and bigotry.
Whether it’s the newly released Beatles video making the fans happy or impersonator deepfake voices that can manipulate people, the positives and negatives of AI is here to stay in the long haul.
A simple command that can generate credible photos or videos instantly threatens to destabilize our notions of what is true or fake.
Fake photos and videos can end up with consequences like couples quarreling over non-existent photos.
World leaders presented with fake war videos can end up making wrong decisions when reality becomes difficult to disentangle from the fakes AI produces.
Human values and preventive measures to avoid such instances attain supreme importance. AI needs a framework of checks and balances which humankind has created in the form of a democratic system, limits to power, and decentralization of decision-making. These systems were devised to prevent the excesses of mankind. AI ecosystem too needs such systems as AI develops so fast to mimic human voice, looks, and thought patterns. As we are building AI exponentially, we need to ensure defensive tech is also built to ensure AI remains transparent and the good causes defeat the bad ones.
At the end of the day, technology is agnostic to ethical values. It’s a double-edged sword that serves its masters in the manner it is programmed to give the output. Nuclear technology which can destroy an enormous proportion in a warhead is the same science that also provides comfort to the world in the form of nuclear energy powering our home appliances. AI is a deep opportunity and trouble depending on how it takes off from here. The builders need to ensure they are building tech responsibly rather than waiting on regulators to regulate it.
Ready to learn more about how Authentify can help enterprises with deepfake detection.
Contact us today for a demo.
WEBSITE: www.pi-labs.ai
EMAIL : hello@pi-labs.ai
Together, let's build a digital future where we can differentiate fake from real.
