Today in the Business of Podcasting
Spotify’s podcast and audiobook discovery will get a boost from Google Cloud’s AI by Amrita Khalid
Spotify is attempting to alleviate the age-old discovery problem in podcasting, as well as audiobooks, with a huge computer. Google Cloud’s large language models are being trained on the roughly 5 million podcasts and 350,000 audiobooks in Spotify’s content library in hopes of finding ways to “augment” metadata to make them easier to recommend. The AI tools in Google Cloud are also expected to help understand patterns in listeners’ favorite spoken content. While Spotify has not gone into detail, there is mention of the Google Cloud AI potentially being used to identify “harmful” content. [Source]
YouTube intros new AI governance policies. AI helps enforce them by Brad Hill
YouTube has recently announced a new initiative to deal with videos altered or generated wholesale by deepfake/AI-powered tools. A new label has been created that will denote a video is “artificial or synthetic” in nature in the description panel. If the topic of the video is particularly sensitive, the label will appear on the video itself. If a synthetically-created video is deemed to be harmful, such as one that shows realistic violence for the purpose of shocking or disgusting viewers, it will be removed outright. YouTube has also promised a future update to its reporting system that will allow users to report a video that is using AI-generated content to imitate “identifiable individuals.” Something for podcasting to keep in mind as the industry grows its presence on YouTube. With it becoming steadily easier and more affordable to access apps that can fully synthesize audio, the question is now raised: will we see similar anti-deepfake policies put into place on podcast apps? [Source]
The case for and against brand safety by Seb Joseph and Krystal Scanlon
Earlier this week long-running women’s news and cultural commentary publication Jezebel was shuttered, and former staff cite brand safety concerns as the reason parent company G/O decided to halt publication. As has been covered many times on The Download, brand safety and suitability measures have a complicated relationship with journalistic content, as covering the news invariably requires discussing topics that flag keyword block lists or make overcautious advertisers nervous. The article offers several counter-arguments to taking an all-or-nothing approach to brand safety measures, including the fact that limiting where one can buy advertising will drive up prices as fewer and fewer publications are deemed safe. Advertising gets more expensive for the brands, and popular long-running publications like Jezebel end up an advertising ghost town. Podcasting need not have this problem. As has been shown in Sounds Profitable’s study Safe and Sound, audiences not only tend to have negative associations with brands running ads on shows with unsavory topics, they tend to have more positive associations with the brand for enabling the podcast to continue production. [Source]
Defining Premium Video – How VAB and Comcast’s FreeWheel are Trying to Settle TV Industry Debate by Jack Neff
The Video Advertising Bureau, representing TV-centric media companies, and Comcast’s FreeWheel ad tech division have teamed up to propose a definition for the term ‘premium video’ with a definitive checklist for buying premium video. The proposal also aims to bring transparency and accountability to measurement and account verification, including full brand safety disclosures and day part log level disclosures to buyers. Clear, fair standards are vital to the growth and health of an industry. Podcasting has made progress on this front and can continue to do so. [Source]
…as for the rest of the news: PodPod covers how to hit the mark with GenZ podcast listeners, and Triton’s newest batch of podcast rankers have arrived, including Australia’s top 100 showing Hamish & Andy has held its #1 spot for six months running.
