No, a deep fake isn’t a football play, but it probably should be. It could be a painting, an internet story or post, or a social media influence. Deep fakes use AI to simulate real images of persons or events to create fake news stories and misleading social media audio and video posts. Deep fake media can get people into trouble, create lawsuits and social disputes and even influence politics.
Deep fakes literally circumvent reality and porport events that didn’t happen. Increasingly, these AI generated images and stories are permeating our society to a point where people don’t know anymore what to believe. “Fake news” is more than the media, politicians and government lying to gain political advantages and control of people, or cover for ill conceived wealth, or to spin stories in a favorable direction.
It literally is aimed at getting people to be confused about real issues, and cause the element of conspiracy theories, and distention, anger and even violence amongst readers. Literally the truth becomes lies and lies become truth. It is so pervasive in our society now that a person finds having to research every post or story to determine what is reality and what is not.
Let’s talk about what a deep fake is and how it gets that name. Deep fakes, or “deep learning algorithyms” teach themselves to crack problems from large data sets from thousands of existing pictures to create a false but realistic media content or images of people.
Once the program collects relevant information from multiple angles, it then maps the image onto another facial image finding commonalities between them. To fine tune the image and make it less distinguishable from the original, another type of machine learning is employed called (GANs) or Generative Adversarial Networks. Through multiple scanning and passes over the images GANs finds flaws and makes improvements until it is very difficult to tell the real from the fake.
Deep fakes have a notoriety for being used exclusively in the commission of crimes. “Sextortion” is a common use, you may have heard of revenge porn. Numerous depictions of famous people have been grafted into pornography settings to extort money and or disqualify their targets from political positions, gain leverage in high profile court cases or smear an opponent. It has even been used to emulate presidential speeches and in official proceedings to skew opinions and damage political reputations.
Deep fake technology is relatively easy to obtain. There are a number of software programs that can make an incredibly complex process easy for the beginner. Programs like DeepFaceLab and FaceSwap are just the tip of the iceberg in AI software technology to begin your career, probably in crime. It was reported in 2019 that 96 percent of deep fakes were of a pornographic nature. But there are organizations utilizing the technology for good also.
Deep fakes are being extensively used in filmmaking. Disruptions in multi-million dollar productions sometimes occur when actors are injured, have serious illness and even die. Deep fakes can aid in recreating facial features transposed onto stand-in actors to create continuity in production and save costs. Examples of this are the “Fast and Furious Series 7” and “Blade Runner 2049.”
Each film used deep fakes to recreate the likenesses of either the main actor in one case who died in a horrific car accident, and the other an actress that was aged beyond the original films timeline.
So, in effect deep fakes are useful in these situations whereby creators can fill in the blanks of lost characters, depict younger versions of actors for flashbacks, and also create wholly new content saving thousands of dollars in having real actors travel to filming sites and in crew costs for filming. I can see deep fakes displacing the need for super-star actors and possibly endangering these high paid careers.
After all, money talks, and it is what drives the ability to make movies. Cost cutting the human element for the AI and deep fake insertions in films could be a move of the future to control or cheapen production costs.
The industry could move from the high paid actors to high paid AI and deep fake technologists that are specifically skilled in creating this content. This technology is also attractive to young programmers and digital film makers. I predict that it will progress into completely studio based films with a lesser need to travel to foreign sets and elaborate outdoor venues.
If the majority of work can be accomplished within the bounds of a computer screen and in studio sets, more small format film makers will latch onto the opportunity to leverage this technology to beat out the big guys.
While deep fake technology may take a chunk out of the movie making process, we are still humans and enjoy the reality of live programming and TV shows. We will likely always embrace the acting talents of certain people, as long as they can keep their mouths about them on their political views.
Acting is an art form and is greatly appreciated knowing that it is real and we know that we are viewing a person we can identify with on either a personal or professional level. It’s hard to identify with a machine generated character. I could never watch the “Fast and Furious 7” scenes where they deep faked Paul Walker into the movie without also knowing that it is fake.
Is it possible that we were “deep faked” on the moon landing? I don’t think so, but some conspiracists will tell you that the whole thing was filmed in a NASA studio or on a back lot behind the Johnson Space Center. To those, I’d have to say, “come out of the basement” every once in a while.
We may never know what is a deep fake and what is not in a movie, but unless it’s depicting some historical content or bibliographical story, who cares. It is all about entertainment. A good portion of the stuff you watch is fake anyway. Fake storylines, fake characters, and yes, fake news.
Learning to discern what is real and fake is becoming more and more difficult, but with a keen eye and a good grasp on reality you usually can tell the real thing from a fake. FBI agents trained in counterfeit bill detecting don’t study the fake bills, they study the real thing so that it is easier to spot the fake one when they see it.
The same should apply to deep fakes. If it seems a facial expression isn’t right or a character’s movements are not natural then it is probably a fake. In other words if it’s too good to be true, it probably is. There is no trick under the sun that hasn’t been tried already. Study history and you will find the same patterns in people over the entire existence of our world.
You might wonder. How do I spot a deep fake? As they get better and better at tricking our minds with all kinds of technological apparatus, I believe that you will still be able to spot deep fakes no matter how advanced they get. A lot of times it’s common sense, a good eye, and observing the original. Look for original sources of photos, video and audio recordings.
We live in a digital world where everything is stored out there somewhere. Unless the FBI, CIA or some other government entity has had this information intentionally scrubbed from the internet or destroyed (a popular ploy amongst some politicians) then it is out there somewhere. Whoever created the deep fake used an original of something that can be found on the internet.
The “mouth” usually shouts out a deep fake. Take notice of audio attempted to match what the mouth is saying. It seems to be the hardest part of creating a deep fake to master, mouth movement to audio.
Just looking at the deep fake without comparing it to an original might not give you many clues, but skin tone, wrinkles, facial perimeters, coloration differences from face to neck or other visible body parts are all good indications that what you are viewing is likely created on the keypad of a deep fake instigator.
And I say instigator because that is what they are trying to do is instigate a problem, and cause a flurry of discussion, conversation, and contention with the viewer. If it is for entertainment value, righteous creators will give a disclaimer to that effect.
To be certain, you can do reverse image searches. I’m not going to go into detail about that, however you can download programs that will perform these searches from a still image that you have screenshot and you can determine from the origin of the photo if it has made the circle of discussion about it being used in deep fake creation. You can gain more context as to how the photo has been previously used, how many times it has been posted, etc.
Deep fakes can be fun or dangerous. Increasingly they are being used for more nefarious purposes. Our political scene is littered with this type of media attacking candidates, current political figures, governments, supreme court Justices and other governing structures to the point of eroding our country’s efficacy in running the government to the will of the people. Be mindful of this technology, don’t believe everything you see.
Question everything, and by all means, be fair in judging what is real or not, as for the sake of truth and liberty, we count on that.
Jon Armour is a contributing editor/author to the line of Design and Construction publications and has 35 years of combined experience across the construction, real estate, and IT Infrastructure industry. He is certified Project Management Professional (PMP), Certified Construction Manager, IT Program Manager, and a published author of a popular Western genre fiction novel. He lives in Magnolia, Texas.
