I can’t stand how much I enjoy Riverside’s AI-powered ‘Rewind’ feature for podcasters

Riverside Introduces “Rewind” Year-End Recap for Podcasters

Riverside, an online platform for recording podcasts, has launched its own take on the annual review trend, similar to Spotify’s “Wrapped.” Their version, called “Rewind,” generates three personalized highlight videos for podcast creators.

Rather than focusing on metrics like total recording time or episode count, Riverside’s recap features a 15-second montage of laughter, rapidly cycling through moments where my co-host and I crack each other up. Another video compiles a series of clips where we repeatedly say “umm.”

Riverside also uses AI to analyze transcripts of your recordings, identifying the word you used most frequently—presumably filtering out common words like “and” or “the.”

Ironically, on my podcast about internet culture, the word “book” topped our list, likely influenced by our exclusive “book club” episodes for subscribers and the frequent promotion of my co-host’s upcoming book.

On another show in our podcast network, Spirits, the most-used word was “Amanda”—not due to any obsession, but simply because one of the hosts shares that name.

We shared our Rewind videos in the network’s Slack channel. There’s an undeniable humor in watching a supercut of people saying “umm” repeatedly. Still, these videos also highlight a growing trend: creative tools are increasingly saturated with AI-driven features, many of which feel unnecessary. The Riverside Rewind is a prime example—while amusing, a video of us repeating “book” offers little real value beyond a quick laugh.

View this post on Instagram

Although I found Riverside’s AI-generated recap entertaining, its release coincides with a period where many colleagues in the industry are losing opportunities to create, edit, and produce new podcasts—ironically, due to the same AI technologies behind these recaps. While AI can automate certain tasks, like removing filler words or silences, the essence of podcasting remains deeply human and creative.

AI excels at quickly producing transcripts, which is invaluable for accessibility and saves significant time. However, it falls short when it comes to making nuanced editorial decisions—such as knowing when a digression is entertaining or when it should be cut for being dull. Unlike skilled human editors, AI lacks the judgment needed to shape compelling stories from raw audio or video.

Despite the proliferation of personalized AI audio tools, such as Google’s NotebookLM, there have been notable failures in their use as creative aids.

Recently, The Washington Post began experimenting with AI-generated, personalized news podcasts.

It’s easy to see why this approach might appeal to executives looking to cut costs—automating the labor-intensive process of producing daily shows. However, the reality is more complicated. The AI-generated podcasts produced fabricated quotes and factual mistakes, which is especially problematic for a news outlet. Internal tests reported by Semafor revealed that between 68% and 84% of these AI podcasts failed to meet the publication’s standards. This highlights a fundamental misunderstanding of how large language models operate: they generate the most statistically likely response, not necessarily the most accurate or truthful—an especially critical flaw in fast-moving news contexts.

While Riverside’s year-end recap is a fun and creative product, it also serves as a reminder: AI is making its way into every field, including podcasting. As companies rush to adopt new technologies during this “AI boom,” it’s crucial to discern when AI genuinely enhances our work and when it simply adds unnecessary clutter.