Riverside.fm's 'Rewind' AI Tool: A Love-Hate Relationship

Riverside’s “Rewind” and the Growing AI Presence in Podcasting

The podcast recording platform, Riverside, has launched its own year-end review feature, mirroring Spotify’s “Wrapped.” This feature, known as “Rewind,” generates three unique video summaries specifically for podcasters.

A Recap Focused on Moments, Not Metrics

Rather than presenting typical statistics like recording time or episode count, Riverside’s approach is different. The first video created is a 15-second montage of laughter, showcasing rapid cuts of moments where podcasters and their guests share a laugh.

A second video follows a similar format, but focuses on instances of hesitation, compiling clips of speakers saying “umm” repeatedly.

Identifying Key Words Through AI Analysis

Riverside’s system then analyzes the AI-generated transcripts of podcast recordings. This analysis identifies the single word spoken most frequently by the participants – excluding common words like “and” or “the.”

Interestingly, for a podcast centered on internet culture, the most frequently used word was “book.” This result was likely influenced by subscriber-exclusive “book club” segments, as well as ongoing promotion of a co-host’s upcoming book.

Another podcast within the network, “Spirits,” found that “Amanda” was their most-used word. This wasn’t due to personal obsession, but rather the presence of a host named Amanda on the show.

The Irony of AI-Driven Features

Sharing these “Rewind” videos within the podcast network sparked amusement. However, it also highlighted a growing concern: the increasing saturation of AI features in creative tools, many of which are unnecessary.

The Riverside Rewind exemplifies the potential uselessness of some AI applications – questioning the need for a video compilation of repeated words. While providing a momentary chuckle, it lacks substantial value.

AI’s Impact on Podcast Creation

Despite enjoying Riverside’s AI recap, its release coincides with a period where opportunities for podcast creation, editing, and production are diminishing for industry professionals. This is directly linked to the same AI tools that powered the Rewind feature.

While AI can automate tasks like removing filler words and silences, the core of podcasting remains a fundamentally human endeavor.

The Value of Human Editorial Judgment

AI excels at generating transcripts, enhancing accessibility and automating a previously time-consuming process. However, it falls short when it comes to crucial editorial decisions.

Unlike experienced human editors, AI cannot discern when a tangential conversation is humorous or when it should be removed due to lack of engagement. It lacks the nuanced understanding required for effective storytelling through audio and video.

Failures in AI-Powered Content Creation

The limitations of AI as a creative tool have been recently demonstrated by high-profile failures, even with advanced tools like Google’s NotebookLM.

The Washington Post’s AI Podcast Experiment

Last week, The Washington Post began testing personalized, AI-generated podcasts delivering daily news updates.

This initiative, driven by a desire for increased profitability, aimed to automate the intensive process of research, recording, editing, and distribution. However, the experiment proved unsuccessful.

Factual Errors and a Lack of Reliability

The AI-generated podcasts contained fabricated quotes and factual inaccuracies, posing a significant risk to the news organization’s credibility. Internal testing revealed that 68% to 84% of the AI podcasts failed to meet the publication’s standards, as reported by Semafor.

This outcome highlights a fundamental misunderstanding of how Large Language Models (LLMs) operate. LLMs are designed to generate statistically probable outputs, not necessarily truthful ones, particularly in the context of rapidly evolving news events. They cannot reliably distinguish between fact and fiction.

A Reminder of AI’s Role

Riverside’s engaging year-end product serves as a crucial reminder. AI is rapidly integrating into various industries, including podcasting.

During this period of rapid “AI boom,” it’s essential to critically evaluate when AI genuinely enhances our work and when it simply produces superfluous content.