The net podcast recording platform Riverside got here out with its personal model of a year-end overview like Spotify’s “Wrapped.” The recap, known as “Rewind,” creates three customized movies for podcasters.
As a substitute of sharing statistics like what number of minutes you recorded, or what number of episodes you made, Riverside created a 15-second collage of laughter, displaying a fast succession of clips by which my podcast co-host and I make one another crack up. The subsequent video is analogous, besides that it’s a supercut of us saying “umm” time and again.
Then, Riverside scans its AI-generated transcripts of your recordings to search out what single phrase you stated greater than some other (we’re assuming they lower out phrases like “and” or “the”).
It’s a bit ironic, however on my podcast about web tradition, my co-host and I stated “guide” extra usually than some other phrase (this was most likely skewed by our subscriber-only “guide membership” recordings… or the truth that my co-host has a guide popping out, which we plug incessantly).
One other present on our podcast community, Spirits, stated “Amanda” extra usually than some other phrase (not as a result of they’re obsessive about me, however as a result of additionally they have a number named Amanda).
Within the podcast community’s Slack, we exchanged our Rewind movies. There’s one thing inherently humorous a couple of video of individuals saying “umm” time and again. However we additionally know what these movies symbolize: our inventive instruments have gotten extra saturated with AI options, lots of which we don’t want or want. The Riverside Rewind factors to the uselessness of those instruments themselves — why would I would like a video of my co-host and I saying the phrase “guide” time and again? It’s good for a fast giggle, however there’s no substance.
Although I loved Riverside’s AI recap, its arrival comes at a time when my trade friends are dropping alternatives to create, edit, and produce new podcasts, due to the identical AI instruments that generated our Rewind movies. However whereas AI permits us to automate away some duties — like modifying out our “umms” and lifeless air — podcasting itself isn’t that mechanical.
Techcrunch occasion
San Francisco
|
October 13-15, 2026
AI can rapidly generate a transcript of my podcast, which is vital for accessibility causes, serving to to automate an exercise that was once extremely time-consuming and tedious. Nevertheless, AI isn’t capable of make editorial decisions round find out how to maneuver audio or video to inform a narrative successfully. Not like the human editors I work with, AI can’t decide when a tangential dialog in a podcast is humorous and when it must be lower as a result of it’s boring.
Regardless of the rise of personalised AI audio instruments, like Google’s NotebookLM, its capability to function a creation software has additionally seen high-profile failures, as of late.
Final week, The Washington Publish began to roll out personalised, AI-generated podcasts in regards to the information of the day.
You’ll be able to see why this would appear like a “good” concept to profit-hungry execs — as an alternative of paying a workforce to do the intensive work of researching, recording, modifying, and distributing a each day present, you could possibly automate it — besides you’ll be able to’t.
The podcasts spouted made-up quotes and factual errors, which is existentially harmful for a information group. In response to Semafor, the Publish’s inside testing discovered that between 68% and 84% of the AI podcasts failed to fulfill the publication’s requirements. This looks as if a basic misinterpretation of how LLMs work. You’ll be able to’t prepare an LLM to differentiate actuality from fiction as a result of it’s designed to supply essentially the most statistically possible output to a immediate, which isn’t all the time essentially the most truthful output — particularly in breaking information.
Riverside did an incredible job making a enjoyable end-of-year product, but it surely’s additionally a reminder. AI is infiltrating each trade, together with podcasting. However on this second of the “AI growth,” as firms tinker with new know-how, we’d like to have the ability to distinguish between when AI serves us, and when it’s fodder for ineffective slop.
