Ever since Alan Turing created a computer that could think for itself in a game of chess around 1950, the advances in artificial intelligence (AI) have been nothing short of amazing. Our society has grown accustomed to asking Siri or Alexa for directions or to perform tasks (an example of AI recognizing and responding to our voices). We are starting to see more and more self-driving cars (AI recognizing the markings on the streets and surrounding objects and making informed decisions). And we are being introduced to a new direction of AI – the ability to make the audience believe that someone is saying something that they are not, in videos known as “deep fakes.”
In deep fake videos, the creator has uploaded hundreds or thousands of images and voice recordings of a celebrity in order for the AI to understand facial structure and voice patterns. Then the person uses a video camera to capture their own face providing the message, and the AI mimics those actions using the celebrity presentation. This August 2022 article in Business Insider explains the origins of deep fakes, features some links to very believable social media clips, and provides keys to detecting what could be a deep fake.
But are all deep fakes bad? There has been some controversy over the use of holograms for music concerts, such as “An Evening With Whitney” which is currently at the Harrah’s Las Vegas. The show’s creators, with the approval of the Whitney Houston estate, have used her likeness to recreate shows which entertain fans that once saw her perform live, and introduce her to a new younger audience – something that would not have been possible after her death. By reading the comments in social media posts, you will find that many people that saw the show were amazed and thrilled with it, while others deemed the idea itself to be “creepy.”
Personally, I think that this type of deep fake can be beneficial, as long as the creator is up front about the use and not trying to be deceitful. One fantastic example took place this fall on the television talent show, America’s Got Talent.
The company is called Metaphysic, and its performers were not behind the scenes. They were in the middle of the stage, actually singing and dancing for the audience to see. But artificial intelligence was picking up their movements on camera and projecting mimicked images of Elvis Presley, Simon Cowell, Heidi Klum and Sofia Vergara – while the latter three (the show’s judges) were sitting in front of them watching the show!
Done in this manner, and not as a means to deceive the viewer, I don’t see any reason why AI-created performances shouldn’t be applauded for the beneficial entertainment value.
