AI v.s. Human: Who’s the better video editor?
Technology has made incredible advances, leading to the development of computational video editing. There are apps available on smartphones these days that can process and edit videos faster than humans.
But the question arises: is an AI video editor better than a human?
While this analysis is arguably subjective, I strongly believe that a human editor is still the better option. But before getting into my reasons, let me introduce a little experiment built to test the hypothesis.
The experiment parameters
First and foremost, AI editing in 2023 is primarily useful for generating short form highlights reel styled content. I’m sure in the future that will change. But for the time being, in order to ensure an even playing field—I decided to limit this test to a ~60 second highlights reel output.
Second, I needed to find a selection of themed video content as a source for the AI and human editor to face off with. Ideally the same assets would be available for both parties.
Then something clicked. A faint memory from a few years ago. I had gone on a work trip to the Cayman Islands back in 2019. Afterward, for fun, I cut together a short highlights reel from a variety of the tropical b-roll I had filmed. Not only that, but I had clipped out and rendered a catered selection from the library of source content that I used when constructing that edit in order to illustrate a media melt package.
What a pleasant surprise—I happened to have a perfect set of assets ready to go for this test. Along with a human edited highlights reel sourced from the same selection of video content.
So after feeding the Cayman Islands video clips into the AI edit software, let’s take a look at the results!
Human editor results
A.I. Editor results
So… Which edit is better?
From my admittedly biased perspective, I would suggest that the human edit is the better result.
Human Editors Win! Yay!!!
^I get to declare this because this is my blog post.
Okay, okay… To be fair, since this is a subjective matter: ultimately the best edit is up for you to decide. But I will try to persuade you.
Thoughts behind my conclusion
Shot selection and structure: Human editors possess creativity and intuition that AI lacks. Human editors can understand the nuances of storytelling that happen when sequencing a string of clips together, and humans can make informed cutting decisions based on how the content will be interpreted by other humans. This ability is something that AI has not yet mastered. And perhaps never will.
Looking at the AI edit, I notice some oddities:
In one clip, poor trimming was at play in the AI edit. The timing of an airplane flying behind a building on the horizon shows a few brief frames of the tail of the airplane. Compared to the human trimming of the same clip, the airplane as a subject is effectively lost in the AI edit.
Airplane clip trimming: AI vs Human side-by-side
Throughout the AI edit there are some redundant sequencing decisions. Back-to-back similarly framed shots were chosen: plants growing on the beach, ferry boats, and cruise ship aerials. The inclusion of these extra takes on the same subject matter makes the whole structure of the edit feel less polished. Compare that to the human edit, where individual takes were selected based on what fit best in the contextual flow of content, opting to not include any similarly framed content.
Redundancy in AI shot selection
Sound design: is an essential component of video editing, and as far as I’m aware—there isn’t any AI tool built to analyze and implement sound design complimentary to video imagery in an edit.
For now, humans have the advantage of using sound to enhance the immersion of a video edit.
In this comparison, the AI has no sound design beyond the music that was selected. So I’m just looking at the human edit to highlight examples where the sound design adds to audience immersion:
You see ocean waves, you hear ocean waves.
You see a man with a camera, you hear the shutter sound when he takes a picture.
You see an airplane flying by, you hear an airplane engine off in the distance.
You see someone walking on the beach, you hear footsteps in the sand.
The human touch is something that can influence an edit in ways that are simply hard to quantify. A video that has been edited by a human will have a unique style and personality.
Human creativity, intuition, and sound design manifest themselves into the core DNA of a video edit, giving it unique traits that set it apart from other videos—even other videos edited by other humans. There will be a small piece of the editor represented in subtle ways on any edit timeline that is constructed by a human. That isn’t always a good thing. But it is something that AI video edits cannot replicate.
However, for all of the negative points I’ve shared against AI video edits here, there are some positive things to consider.
Use cases for AI video editing
AI video editing is fast and efficient, and it can handle large volumes of video content with ease. That fact gives AI editing some solid real-world use cases…
Easily generate a ton of themed video content to fill out a social media posting calendar.
If shooting a video that involves a lot of b-roll—you can send a rapidly cut, low stakes highlights reel to your client immediately following a production day. Simply to share a feel for the content captured and build anticipation for the final product down the road.
Use AI video edits to create memories from home video captured on your cell phone. This is one application I actually use a lot. It’s a fun and easy way to celebrate and remember adventures with your loved ones. As a bonus, most smartphones these days are automatically doing this for you.
