I Love to Hate Riverside’s AI “Rewind” for Podcasters

The podcast recording platform Riverside recently launched its own year-end review feature, mirroring the popular concept of Spotify Wrapped. Dubbed “Rewind,” this tool generates three personalized video clips for creators, offering a quirky, AI-driven snapshot of their year in audio. Rather than presenting dry statistics like total recording minutes, it focuses on the human moments, and verbal tics, that define a show’s character. This playful feature highlights how deeply artificial intelligence is weaving itself into the creative process, for better or worse.

My own Rewind experience was both amusing and ironic. The first video was a fifteen-second collage of laughter, a rapid-fire montage of moments where my co-host and I made each other crack up. The second was a similar supercut, but this time it compiled every instance of us saying “umm” throughout the year. The final clip used AI-generated transcripts to identify our most frequently spoken word, presumably filtering out common articles and conjunctions. For a podcast centered on internet culture, our winning word was “book.” This was likely influenced by our subscriber-exclusive book club segments and the relentless promotion of my co-host’s upcoming publication.

Across our podcast network’s Slack channel, creators shared their Rewind videos. There’s an inherent comedy in watching someone repeatedly utter a filler word, and the clips sparked genuine laughter. However, the exchange also prompted a more sobering reflection. These videos symbolize the increasing saturation of our creative tools with AI features, many of which feel superfluous. What substantive value does a loop of someone saying “book” actually provide? It’s a momentary gag, not a meaningful insight. The feature feels emblematic of a broader trend: the proliferation of AI-powered tools that prioritize novelty over utility.

This arrives at a precarious time for the industry. While I enjoyed Riverside’s lighthearted recap, many of my peers are confronting a harsh reality. The same technology that automates transcriptions and creates fun year-end compilations is also being leveraged to replace human roles in creation, editing, and production. AI excels at automating tedious tasks, like removing dead air or generating a transcript for accessibility, work that is invaluable for saving time. Yet, podcasting is an art form, not a mechanical process. AI cannot replicate the nuanced editorial judgment of a human producer who knows when a tangential anecdote is gold and when it’s merely self-indulgent filler.

The limitations of AI in creative arenas have been starkly illustrated by recent high-profile stumbles. Last week, The Washington Post began offering AI-generated, personalized news podcasts. To profit-driven executives, this likely seemed like a brilliant cost-cutting measure: eliminate the need for researchers, reporters, editors, and sound engineers. The result, however, was a failure. The podcasts generated fabricated quotes and contained factual errors, with internal reports indicating that a staggering 68% to 84% failed to meet the publication’s basic standards. This debacle points to a fundamental flaw in applying large language models to journalism. These systems are designed to predict the most statistically likely response, not to discern objective truth, a dangerous shortcoming for any news organization.

Riverside’s Rewind is a clever and entertaining product. It successfully creates a shareable, personal moment for creators. But it also serves as a pointed reminder. As the AI boom accelerates, with companies eagerly integrating new technology into every facet of work, we must develop a critical eye. The key is distinguishing between AI that genuinely serves a purpose and AI that merely generates digital clutter. For now, the heart of podcasting, the connection, the storytelling, the spontaneous humor, remains firmly, and thankfully, a human endeavor.

(Source: TechCrunch)