Who’s better at selecting and editing talking head clips from a video podcast: Rooster High’s editors (like founder Zach) or an AI tool?
The AI clip-finding tool used in this test is part of Descript ↗️.
AI Tools in Content Marketing
There are plenty of places that Large Language Models ↗️ and their associated applications (commonly called AI, or Artificial Intelligence) boost a creative process. Generating ideas, reviewing written words, or summarizing large volumes of text have high rates of usability. Like any new tool, we have to look at the context in which we’d use it, test it, and consider the ramifications of the use of the tech on our process.
Talking Head Clips
Rooster High’s content marketing offers for service professionals often ends up including a talking head video of some kind, often clips pulled from a video podcast. In creating a content strategy that serves real audience relationships, we’ll often recommend the video podcast to clips for social approach for anyone who needs to show their face and voice their audience while demonstrating their expertise to build trust.
Rooster High spends a lot of time pulling clips from video podcasts. Those clips need to go from living in the middle of an episode to being a good short-form video for social media. The TikTok or Instagram algorithms aren’t going to give us a break just because it’s a podcast. We need the content to transform from “quote in the middle of an episode” to “attractive short-form content” using only our editing skills, and while showing off the best, authentic parts of our subject.
The Test
The question is: will Large Language Models and AI tools replace the human in selecting and editing podcast clips?
Human Clips
AI Clips
Click on the videos above to view each batch of clips.
To test this, above are six clips founder Zach pulled from MOM-enomics against AI’s work on the same episode. The episode title is “Tips for 20 Years of a Happy Marriage” where host Booth Parker discusses the ways she and her husband contributed to the success of their relationship.
So who wins?
The AI-selected clips aren’t bad, but let’s look at the real difference between the two sets of videos: all the AI did was exactly what it was supposed to, which was pick clips.
The AI tools only select clips, they don’t edit. Editing is absolutely necessary if you want to turn long-form video podcasts into valuabe short-form content. It takes an empathetic narrative sense to best select and translate clips to short-from.
Additionally, a human selecting clips doesn’t take much time at all if it happens during the editing processes. The human takes many steps, and the AI tool only replaces one of them.
The human-edited clips weren’t just selected. They were edited for speed and pacing, strong hooks to retain viewers in the first three seconds, sometimes the hooks were brought from further into the clip to to front, and with intention to show off the host authentically. And all this was done with a knowledge of the host’s brand, so that she was not misrepresented at all for the sake of more views.
Are AI tools useful for video podcasts?
Absolutely. The tool used for the test outlined here was Descript, which contains a large amount of AI tools. Many of them simply speed up or replace low-discretion tasks we’re already doing in the editing process, and help create and efficient process.
But AI tools will never replace informed human decisions; they are always making a “best guess” at what it thinks a human might want. And rarely is a high volume of machine guesses the right call in a content marketing strategy focused on being one part of the real, human relationships your businesses cultivates to grow.