We’re not even halfway through the year, but 2023 has certainly already become the year of artificial intelligence (or, as it's more commonly known now, AI). From AI-powered products and updates from NAB to new short films and other AI-produced projects hitting the internet, there's so much to follow and keep tabs on — especially if you’re worried about AI eventually taking your job.
While I’d personally argue that AI will most likely be used more for speeding up workflows and reducing grunt work on most productions, there is a looming threat of full AI replacement for different film and video professionals.
One of the first arenas in which AI looks to move into is video editing. But how real is this threat today? More importantly, are there some ways (and some research proving) that AI might not fully replace human editors?
1. Genuine Human Emotion
According to an article published by MIT, AI is attempting to tackle human emotion as we speak. As the journal states, “[W]hile humans might currently have the upper hand on reading emotions, machines are gaining ground using their own strengths.”
However, unlike other forms of AI where lots of information can be gathered very quickly, AI is struggling to keep up with emotion simply because of the lack of large amounts of data — most notably in terms of machine learning of human voice and face cues which are some of the only indicators of human emotion which machines can even attempt to track.
Others, like Bimo Putro Tristianto, have argued that “despite all of these advances, it’s important to note that AI systems are still far from being able to feel emotions in the same way that humans do. While they may be able to recognize and respond to emotions, they do not have the same complex psychological and physiological responses that humans do."
This means that today, and possibly for quite some time in the future, the idea of AI being able to replicate genuine human emotion and therefore being able to use that information to make complicated decisions in the video editing process to convey these emotions doesn’t seem very likely.
'Call Me By Your Name'Credit: Sony Pictures Classics
2. Dramatic and Comedic Beats
Using some of the same logic from above, without the same sense of genuine human emotion at its heart, AI will also always struggle with others recreating these human emotions in an edit for both dramatic and comedic moments. In particular, I’d personally argue that comedy might be the hardest thing for AI to ever replicate in any true sense.
Even for humans, comedy is hard and very subjective from one person to the next. Yet, as a collective audience, we do know funny when we see it. Humor comes from a deep understanding of the human experience and requires pinpoint control to subvert expectations, challenge social norms, and simply set up jokes with perfectly timed punchlines.
3. Ironic Moments
Along with comedy, irony might be hard to explain to another human learning about it for the first time, much less to a machine or algorithm. Irony, in particular, is laced in human experience and might be quite hard (if not impossible) to connect to a certain facial cue or any other data point that AI could even track.
In the edit, irony is quite hard to create, but seasoned editors make comedic or dramatic decisions all the time, which help tell a story with these elements that AI might not be able to create on its own anytime soon.
4. Free Association and Improv
The same goes for anything free-association or improv-based. In a way, improv itself is its own form of human algorithmic performance. A person takes a prompt and puts it through their own human experience and delivers a presentation of the same information as before, but changed based on who they are and their own creative process.
For video editors, there are a lot more elements of free association and improv than you might think too. Since there’s obviously never one path to take when taking footage from a shot and putting it together into a video, the amount of loose improvisation an editor must do is nearly endless. While AI can make many similar decisions, its complex algorithm will never be more unique (or specifically more human) than yours.
5. Natural Timing
Finally, and this one might be a bit debatable, natural timing is something that AI will most probably always struggle with, and timing is a huge part of video editing. Yet, if you had to describe to another person learning editing for the first time, it would be hard to explain what a “beat” is and how they should choose to create one in an edit.
There are certain hard fast rules to the number of frames before or after a cut to consider, but even the most rigid of video editors would tell you that all those rules are up to interpretation and could always be changed based on an editor's feel for the natural timing of the piece.
From having messed around with plenty of scenes with intricate timing in a sequence, the differences between one or two seconds at a time can vastly change the feeling of a scene and how it can make an audience feel.
But what do we know? For all you know, this article was written by AI.
So, who’s to say what the future actually holds? Still, let us know your unique human thoughts in the comments below as to how you see AI playing out in the world of video editing.
