I have a love-hate relationship with Riverside's AI-powered 'Rewind' feature for podcasters. It's a fun, light-hearted take on the year-end review concept, but it also serves as a stark reminder of the growing influence of AI in our industry.
Riverside, an online podcast recording platform, has created its own version of a year-end recap, similar to Spotify's 'Wrapped'. Their 'Rewind' feature generates three unique videos for podcasters, showcasing a different side of their content.
The first video is a hilarious collage of laughter, a rapid-fire sequence of clips where my co-host and I crack each other up. The second video is equally amusing, a supercut of us saying 'umm' repeatedly. It's a light-hearted look at our podcast, but it also raises some interesting questions about the role of AI in podcasting.
Then, Riverside's AI-generated transcripts come into play. The platform scans through the recordings to find the single word that you've said most frequently (excluding common words like 'and' or 'the'). In my case, it was 'book', which is ironic given that my podcast is about internet culture. This was likely influenced by our subscriber-only 'book club' recordings or the fact that my co-host has an upcoming book release.
Another podcast in our network, 'Spirits', had a similar experience. Their most frequently used word was 'Amanda', not because they're obsessed with me, but because they also have a host named Amanda!
When we shared our 'Rewind' videos in the podcast network's Slack, it sparked a mix of laughter and contemplation. While it's entertaining to see these videos, they also represent a shift towards AI-saturated creative tools. We're at a point where our industry is losing opportunities for human creativity and craftsmanship due to AI automation.
AI can certainly streamline certain tasks, like editing out 'umms' and dead air, but podcasting is an art that goes beyond mechanics. It requires human editorial choices to tell compelling stories.
The recent TechCrunch event in San Francisco highlighted this further. AI tools like Google's NotebookLM can generate transcripts quickly, which is beneficial for accessibility. However, they fall short when it comes to making creative decisions about audio and video editing. Human editors have an intuitive understanding of when a tangential conversation is funny and when it should be cut for the sake of the narrative.
The rise of personalized AI audio tools has also seen its fair share of failures. Take, for instance, The Washington Post's experiment with AI-generated podcasts. The podcasts produced made-up quotes and factual errors, which is a significant concern for a news organization. Internal testing revealed that a staggering 68% to 84% of these AI podcasts failed to meet the publication's standards. This highlights a fundamental misunderstanding of how LLMs work. They are designed to provide statistically probable outputs, not always the most truthful ones, especially in breaking news scenarios.
Riverside's 'Rewind' is a well-executed, fun product, but it serves as a reminder of the challenges we face in this 'AI boom'. As companies experiment with new technologies, we must discern when AI is a helpful tool and when it's just creating useless content.
In this ever-evolving landscape, it's crucial to strike a balance between embracing innovation and preserving the human element in our creative industries.
What are your thoughts on the increasing role of AI in podcasting? Do you think it enhances or detracts from the art of podcasting? I'd love to hear your opinions in the comments!
