Artificial Intelligence is a real thing. The building blocks are readily available to anyone who can afford to rent GPU power and string together Python code. As technology advances and it becomes easier to manipulate the human form with AI, will Hollywood continue to exist? Will this new art form be distributed in a more democratic fashion? Or will Hollywood maintain its grasp on the film industry?
The news cycle is full of stories about AI, particularly since the inauguration of President Trump. AI has become so incredibly convincing, that it’s possible to create fake political ads, pornographic videos featuring “whoever you want”, or even craft a believable hoax.
All art is emotion, and that the task of the filmmaker is to use the tools of his medium to manipulate the audience’s emotional experience. — Alfred Hitchcock
There are moral implications to this technology. Beyond simply creating art, AI can take the likeness of anyone and make them do or say whatever you want. Since technology has become so realistic, unless the final product is clearly satirical, it can potentially be mistaken for the real thing.
The benefits of this technology far outweigh the detriments. AI is the equalizer when it comes to creating film art. One day there will be a featured film created entirely using AI. That film will require no body double for nude scenes, no scheduled time with actors, and no stunt doubles. To see what the future of AI in film looks like, we simply need to compare what animation looked like before Pixar released “Toy Story” and what it looks like now.
What is Deepfake?
Deepfake is the use of machine learning to superimpose images and videos on other images and videos. An example of Deepfake would be to put Alec Baldwin’s face on Donald Trump’s face, recreating the SNL skit. In the skit, Baldwin imitates Trump, in the Deepfake Baldwin is Trump.
Deepfake uses source data and input data to superimpose an image on the other. Rather than a video editor or programmer editing every frame in a clip, a Neural Network does all the heavy lifting. With Deepfake and Neural Networks, time and GPU power are the currency and constraints.
What Is A Neural Network?
A Neural Network is essentially something that recognizes patterns. Modeled after the human brain, a Neural Network can be trained to recognize patterns and make predictions. When patterns are grouped or classified its called supervised learning. When data is uncategorized and unguided, it is considered unsupervised. In the case of Deepfake, we are using supervised learning to train the neural network to superimpose images on other images.
The ability to recognize patterns and make predictions are what makes Neural Networks powerful. A simple example outside of Deepfake would be to train a Network with a data set that contains test scores as well as time spent studying before the test. This information could be used to produce a prediction based on our criteria — like what your grade would be if you studied 3 hours for a test versus 8 hours. Machine learning doesn’t allow us to see into the future, but by turning everything into numbers, AI can see patterns that we don’t and can predict outcomes with a high level of accuracy.
Will Deepfake Destroy Hollywood?
Deepfake has found popularity because of its notoriety. With fake political videos, porn, hoaxes, and the negative press surrounding the technology; most press has focused on the negative applications that are possible with the technology, not the positive possibilities.
For about 100 years Hollywood has controlled the movie making business. In early Hollywood, the sets were regularly recycled as was the talent. In some parts of the word, new films were created entirely by splicing older existing films. We are inching closer to a time where computer power and AI converge to allow the creation of complete full-length features without a cast, or a massive budget.
The essence of cinema is montage, that a film story is best told by cutting between discrete pieces of film.The basic structure as a collision between shot A (“thesis”) and shot B (“antithesis”) to create a completely new idea (“synthesis”) in the mind of the viewer. — Kuleshov
Deepfake is one application of AI. It is troubling when it’s used for evil, and can be hilarious or satirical when used to create art. It presents moral challenges particular in regard to its use in the adult film industry — but it also presents new opportunities for creative artists to have access to a medium they wouldn’t have had access to otherwise.
The question of whether Hollywood will continue to hold its iron grip on the movie industry is moot. We’ve already seen the music industry disentangle. Music can be replicated rather easily if someone has musical ability. To create a unique full-length video — one must currently employ a staff of people, and be skilled in a variety of areas. Deepfake and AI will one day allow the non-technical creatives to create clips and full-length videos using simple tools. We replaced the book with the e-book, and the musician with “indie” music and cable television has been almost eliminated by Youtube, film is the final frontier. It will be the tools built based on AI that will be the equalizer. Video created using computer power and arranged using our human creativity, will disrupt the film industry. AI will change not only the media we consume but also the media we create.