When it comes to streaming, engagement is often key, with services and platform players grappling with ways to not only encourage users to tune in and sign up, but to keep them watching for longer and coming back more frequently.
During a keynote at the recent StreamTV show in Colorado, Allen Media Group President of Digital Michael Senzon spoke about the importance of engagement and cultivating daily viewing habits, but he also dreamed big about how AI will ultimately change not just FAST but streaming as a whole.
Before diving into his vision of an AI-driven future that captivates and engages audiences daily, Cenzon revealed some interesting statistics and advice based on AMG's Local Now current efforts. This free ad-supported streaming TV (FAST) service stands out in a crowded space by choosing to lead by relying on its own technology, human capital, multiple partners, and focusing on niche local programming. In terms of creating daily habits, he said the most valuable ARPU users are typically using the service or platform at least two or three times a week. For services, engaged users are monetized and mean they don't need as much marketing spend. Cenzon said Local Now in particular has seen “exponential growth” in time spent per unique user.
The free, ad-supported streamer counts broadcasts across major platforms and has worked to build a user journey that includes multiple viewing sessions throughout the day: for its most loyal users, viewers first watch exclusive local news stations (there are 223 geo-targeted Local Now channels across the U.S.) for about an hour, then move on to their favorite affiliates like CBS, followed by national news like Bloomberg or Dateline NBC, before leaning into entertainment with single-series IP channels from content partners and, on very successful days, wrapping up with a movie in the evening.
And it's working: With a focus on content and curation, leading with something unique followed by a wide range of news and entertainment options, Senzon said 70% of Local Now's viewers are now loyal users who use the service daily, up from just 15% last year.
Stream GPT?
Senzon certainly wasn't the only one talking about AI on the StreamTV show. As he pointed out, there are some commonly cited use cases for AI streaming that are likely coming soon, such as AI deciding which FAST channels to promote, predicting which movies to put on channels or what viewers should watch, and helping to create FAST channels. But he took the time to envision how AI will change FAST and streaming over the next five years or more, laying out some scenarios.
To provide context, Cenzon pointed to the early 2000s, when visiting multiple web pages a day was a common habit, before social platforms like Facebook came along and became dominant, eager to keep users within their platforms. He felt the industry was “on the brink of that with streaming,” pointing to smart TV makers like Vizio, Roku and Amazon investing in TV OS with the aim of becoming the sole destination for viewers.
“AI will go a step further and do for streaming what social did for the web page,” Cenzon said on stage during the keynote.
But he also believes AI has the potential to transform TV OS and OEMs overall.
Senzon pointed to the current ecosystem where multiple fragmented gateways for services and content partners are integrated in various ways. For example, a service like Local Now can have a direct agreement to be included as a standalone app embedded in a smart TV interface or streaming stick (of which there are many), or it can act as a content partner for a platform or have a deal to have its channel included in a FAST service lineup such as The Roku Channel.
The broader picture is that GenAI's predictive capabilities could even replace OEMs' proprietary operating software five years from now.
“Imagine what would happen if the next version of an OEM's OS was actually 'streamGPT,'” Senzon said.
He described it as a model that understands users, their behavioral patterns, and gets to the heart of what they actually want to see and feel — like when you turn on your TV and it says, “Hello Michael, by the way, we've made a TV show that you might like.”
Outside of streaming, he pointed to tech efforts like the Apple Watch and Elon Musk's brain chip company Neuralink as examples of how AI could potentially connect with emotions and understand what humans want.
Cenzon also asked the audience to imagine an AI-curated FAST channel designed just for the viewer, along with a feed of their favorite TikTok videos, all seamlessly delivered through an AI-powered model that knows what content the user wants to watch.
Of course, many questions can easily arise about such features and characteristics, and Cenzon points out that technology aside, rights and privacy are both aspects that need to be elucidated.
GenAI TV – The end of the content arms dealer?
Furthering the AI-enabled path, Senzon spoke about the potential for GenAI to revolutionize content creation.
Some companies are already working on AI-generated content and TV series. Cenzon cited OpenAI's text-to-video generation tool “Sora,” which will soon be publicly available, as an example. He also mentioned Showrunner, a text-to-episode generation platform from The Simulation, which calls itself the “Netflix of AI.” The company announced its South Park AI episode generation tool last year, and in May launched a production platform that allows users to create their own TV shows using AI. The company announced 10 web series TV shows produced on its Showrunner platform. Based on a 10-15 word prompt, the platform can generate scenes or episodes of 2-16 minutes, including AI dialogue, voice, editing, coherent characters and storyline, according to Forbes. Currently, episodes can be generated in styles such as cutout, anime, and 3D animation (not live action).
In an interview with Forbes, Edward Saatchi, CEO of Simulation (formerly Fable Studios), said that AI in general, and its generative platform in particular, isn't well suited to entertainment series with complete series arcs like: Breaking Bad.
“Currently, AI can't sustain a storyline for more than one episode,” Saatchi told Forbes. “What AI does best is highly episodic shows like sitcoms, police dramas, and space exploration, where the characters pretty much reset every episode.”
Speaking on the StreamTV show, Cenzon hypothesized that an effort like Showrunner could move the needle, posing an intriguing question:
“if [content] Will arms dealers be a problem in the future? The next TV show you watch will be made by AI, Breaking Bad? he argued.
But his dream of AI in streaming may be a long way off, and the executive acknowledged that it will require the alignment of other factors and stakeholders to make it a reality, including Hollywood unions representing actors, writers, and directors, such as the Directors Guild of America, Writers Guild of America, and Screen Actors Guild (SAG-AFTRA), as well as intellectual property rights.
Here are some other fun ideas Senzon proposed for AI-powered streaming (again, keep in mind that these are just ideas, and to make them a reality you need to figure out not only the technology, but also the demand, the rights, and not to mention the potential costs involved):
- The AI-generated NFL game will look like a real pro football event, but will feature the greatest players of all time and air in mid-summer.
- Your favorite movie sequel (in Senzon's case, it was Back to the Future 4)
- Evening news produced especially for you
- User-level curated FAST channels, including programming created specifically for users
- AI-generated presidential debates between past and recent presidents or candidates (e.g., Abraham Lincoln vs. former President Trump)
“If you look at it with an open mind, it's really amazing,” Cenzon said, “and I think that's ultimately why AI will create the ultimate streaming service and make it the ultimate daily habit.”
Developing a personalized recommendation engine
And while acknowledging it's a “lofty idea” that may not come cheap for the future of AI in streaming, Cenzon also pointed to another initiative AMG is already working on as a long-term partner with Google around a personalized recommendation engine for users.
For example, if one user likes romantic comedies and another likes Architectural Digest, one user's FAST EPG will look different from another user's.AMG and Senzon have been talking about personalized FAST channels for at least two years now, including when they partnered with Quickplay on the OTT platform in September 2022.Quickplay announced earlier this year that it would partner with Google to use GenAI for streaming video discovery, enabling content programmers to create dynamic storefronts with personalized recommendations.
While some iterations of similar features are in use today, such as surfacing recommendations based on other titles you've watched, Senzon said the recommendation engine built on the backend of Local Now mentioned in the StreamTV show hasn't been implemented on the front end yet.
