How to Save Money in OTT with AI

Learn more about AI and streaming at Streaming Media East 2022.

Read the complete transcript of this clip:

Nadine Krefetz: Realistically, you're training a system with information to look for certain things. And in Zixi's case, you're training within the transfer side. In Michael's case, it could be on the production side or the monetization side. For Ethan, realistically, if you're not getting cost savings, what kind of value are you seeing? Because now we started to talk about the fact that you can have people doing other things, but Ethan, for instance, are there things that you are having delivered, is there value that you are creating, that you could not create in another way?

Ethan Dreilinger: From our perspective, you can create the value in another way, but to create that value in another way would be far more costly than applying AI to it. And I think if you look at the economies of scale, it's not just an OPEX/CapEx discussion. It's really about user experience. And then ultimately on the advertising side, what the value of that is. So you go back to like the first example that I had where, as far as weather goes, all I want to know is the pollen count, right? All the rest of the stuff is just useful information to me, but pollen count to me is actionable. We can start to target that because now the AI knows that that's of interest to me.

So you can start to send ads to me that are tailored to that. It also makes experience better. So this way, when I go looking for weather, I'm gonna go back to the place that has that information easy for me to find, making it simple and making my life better. Downstream of any AI that was applied, it makes my life better as an end user. So it goes back into the stack. It's the same as what Eric is doing at Zixi, where, on that QoE, if you can continue to deliver great streams of high quality, I'm not gonna fall off of it. I'm not going to say, that stream was awful. I'm not going back to that anymore, deleting that app, because I know I'm gonna get a consistent experience. And I think there's a lot of value in that. It's not value that hits a ledger anywhere, but it's value to the brand and to the end user.

Nadine Krefetz: But would you say you are able to create something with machine learning that you would not be able to create otherwise?

Ethan Dreilinger: Oh, absolutely. And again and again, that goes into that application of AI or machine learning into your particular foundation. and what you're looking to do with it. We've done some projects where we were creating highlights using AI, no humans involved, just creating highlights from different sporting events. We've done some stuff on the political side with residential and congressional debates where you're pulling out the highlights of the debates and sort of matching up the responses. So they kind of make some sense and stuff along those lines. So there's a lot of value to be created there, Is there an audience for that? That's the next question you have to figure out. And then obviously there are some ways to use AI to find the audience for that stuff you've just created. Probably not a great business model, but you can then start to use those AI learnings that you've got. With what Michael's doing with transcriptions, if discovery is the next step from that, let's take those transcriptions, and feed them back into a natural-language understanding model. And then you can start to create derivatives of that programming and start to serve up some bite-sized pieces of programming that drive people back into full content. So there's lots of different ways to start to apply that and create value from it.