The online podcast recording platform Riverside recently released its own version of a year-end review, similar to Spotify’s “Wrapped.” This recap, called “Rewind,” creates three custom videos for podcasters. Instead of sharing basic statistics like recording minutes or episode counts, Riverside focuses on more personal moments. For example, it created a fifteen-second collage of laughter, showing a quick succession of clips where my podcast co-host and I make each other crack up. Another video is a supercut of us saying “umm” repeatedly.
Then, Riverside scans its AI-generated transcripts to identify the single word you said most often, presumably filtering out common words like “and” or “the.” Ironically, on my podcast about internet culture, my co-host and I said “book” more than any other word. This was likely skewed by our subscriber-only “book club” recordings and the fact that my co-host has a book coming out, which we mention frequently. Another show on our podcast network, Spirits, said “Amanda” most often, not because they are obsessed with me, but because they also have a host named Amanda.
In our podcast network’s Slack channel, we exchanged our Rewind videos. There is something inherently funny about a video of people saying “umm” on a loop. But we also recognize what these videos represent: our creative tools are becoming more saturated with AI features, many of which we do not want or need. The Riverside Rewind highlights the potential uselessness of these tools. Why would I need a video of my co-host and I saying the word “book” over and over? It’s good for a quick laugh, but it lacks substance.
Though I enjoyed Riverside’s AI recap, its arrival comes at a time when my industry peers are losing opportunities to create, edit, and produce new podcasts, thanks to the same AI tools that generated our Rewind videos. While AI allows us to automate some tasks, like editing out “umms” and dead air, podcasting itself is not that mechanical. AI can quickly generate a transcript, which is important for accessibility and automates a formerly time-consuming task. However, AI cannot make editorial choices about how to maneuver audio or video to tell a story effectively. Unlike human editors, AI cannot determine when a tangential conversation is funny and when it should be cut for being boring.
Despite the rise of personalized AI audio tools, their ability to serve as creation tools has seen high-profile failures. Last week, The Washington Post began to roll out personalized, AI-generated podcasts about the daily news. You can see why this would seem like a “good” idea to executives, as it could theoretically replace the intensive work of a human team. However, the podcasts spouted made-up quotes and factual errors, which is existentially dangerous for a news organization. Internal testing found that between 68% and 84% of the AI podcasts failed to meet the publication’s standards. This seems like a fundamental misinterpretation of how large language models work. You cannot train an LLM to distinguish reality from fiction because it is designed to provide the most statistically probable output, which is not always the most truthful, especially with breaking news.
Riverside did a great job making a fun end-of-year product, but it is also a reminder. AI is infiltrating every industry, including podcasting. In this moment of the “AI boom,” as companies tinker with new technology, we need to be able to distinguish between when AI truly serves us and when it is merely fodder for useless slop.