5 comments

  • simojo 6 hours ago
    When searching for podcasts that interview researchers I'd like to hear about, I often have to scroll past many AI-ingested arxiv summaries before finding what I'm looking for. It's not inherently bad to have those TTS summaries, but they should reside in their own place rather than the general podcast space.
  • 1123581321 4 hours ago
    Castro (Apple ecosystem podcast client) just de-indexed Inception Point AI spam. Hopefully other clients and eventually major indexes follow suit. Most podcast clients are careful about opining on what their listeners watch and don't curate the index at all, so a removal is significant. https://castro.fm/blog/hiding-inception-point-ai
  • alexgotoi 3 hours ago
    Same thing that happened to blogs and SEO, now hitting podcasts. AI can churn out mediocre content faster than humans can produce good stuff, platforms haven't figured out filtering yet.

    We're probably 2-3 years away from decent AI content detection/filtering. Until then it's just noise. The long-term play is building actual audience relationships so people come directly to you instead of through discovery platforms full of slop.

    I will include this thread in the next issue of https://hackernewsai.com/

  • jajuuka 3 hours ago
    This is a fairly novel use for AI. I would be curious to do a comparison between someone I was very familiar with and their generated voice. The podcasts Google generate are sound but the presenters lack any sense of personality and feel as deep as cardboard.

    Most of my podcast listening though is of panel discussions which you can't really outsource in a meaningful way.

  • theothertimcook 5 hours ago
    It’s been under siege for years with human slop, it used to require a combination of skill, will, and means.

    Podcasting was always doomed to die at the hands of platform inflation, the signal/noise ratio has steadily been worse for years.

    • xnx 4 hours ago
      Yes. Quality of podcast content is just a few years behind the fall in quality of web content. Like the web, you now need an agent to preprocess the miasma of podcast slop into something useful.