Tech focus: Exploring the AI opportunity
Magnifi’s AI tools have been utilised on major cricket tournaments
In the latest tech focus feature, Backlight Streaming, Magnifi, Newsbridge and Veritone look at how AI is already being utilised across sports broadcasting and considers what the future might hold.
Artificial intelligence (AI) is already playing a huge role in sport. From indexing and asset management to highlight generation and player statistics, the use of AI is helping to personalise fan experiences, create additional monetisation opportunities and deliver more content to more platforms, more quickly. And the exciting part is that we’re still in the infancy of what AI can do to transform how sport is delivered, consumed and monetised.
“AI for us is basically a toolbox, a set of technologies, which are made up of computer vision, audio analysis, data meshing and machine learning. Its purpose is essentially in our case, to help our customers be more efficient in their goals, which is basically to create and distribute content, to reach the fans wherever they are on their journey,” says Cristian Livadiotti, co-CEO, Backlight Streaming.
“AI allows broadcasters to produce more content with the same number of people. The big benefit is reducing the cost required to produce content to cover the sports that they have acquired in the best way.”
This idea of using AI to improve productivity and enhance efficiency is picked up on by Newsbridge CEO Philippe Petitpont. “The goal is to have autonomous users so journalists can have more superpowers with AI so that it can work faster and they have everything they need to build a story.”
So how, specifically, is AI being used in sports? Gary Warech, head of sports & entertainment, Veritone, offers an overview: “It all starts with content management, including how content is ingested, tagged, accessed and activated. Content creators can improve their content management workflows through strong AI-powered engines designed to help process and organise massive amounts of content data. From there, they can consider expanding to create new forms of content at scale. AI streamlines content processes for broadcasters and federations to make it easier when working with multiple stakeholders.”
Gary Warech, Veritone
He gives the example of a major golf tournament. “It leverages AI by ingesting all live video and audio from the event site, metadata tags it, archives it, then pushes the feeds back to multiple broadcasters and sponsors simultaneously. As all content is now archived and searchable via AI, they can also license their content around the globe for future revenue opportunities. Licensing can be accomplished either through a self-service platform or through a professional service offering. Viewers benefit from increased production quality and on-air features as well as new and exciting ways to engage with the content whether or not they are physically able to attend events.”
A few highlights
“One of the ways AI is currently being utilised in sports is how we are doing it, which is creating key moments and highlights,” explains Meghna Krishna, CRO, Magnifi. “As the need for personalisation increases and as the need for real-time publishing of content or data to social media increases you need AI to be able to create content faster and publish it to the different social media platforms.”
Livadiotti goes even further, introducing the concept of hyper specialised content. “When we see the digital maturity of our customers, they are now endorsing a 360 digital landscape big time, meaning that they will hyper specialise the content by platform. So they’re hyper diversifying the content they’re producing, targeting the platform for what it brings. And this is key for them to prevail in this hyper fragmented digital landscape. In order for them to have a good performance of digital, they have to be everywhere, but they have to be everywhere in the right way, and use the best of each platform. AI allows them to be efficient and to generate that large variety of content with an appropriate cost.”
Read more Ad tech tools take monetisation and engagement to the next level
Of course, there’s little value in paying for rights and producing reams of content if it’s not generating any revenue, so the monetisation opportunities that AI technology enables is also front of mind for many.
Krishna cites the example of the IPL, one of the largest sports properties in India. “Previously they only sold broadcast rights for a certain amount of money. This year they’ve sold broadcast rights and digital rights separately, and they’ve multiplied the revenue that they generated from it threefold over the last few years. Federations, clubs or conferences can actually further sell the content and get larger virality through the players and increase the popularity of the players. This also helps with a high level of monetisation because now you’re bringing that sport or athlete closer to the fan.”
Content personalisation also has a role to play here, she adds: “If the ads that you’re now offering are more personalised based on the data analytics that you’ve been able to mine, people are more likely to interact with the ads because they’re of actual interest and the monetisation from that point of view also increases.”
Player-specific content is proving popular across many sports, providing a way to engage fans between matches and give them new information, behind the scenes access and even more stats and data.
Petitpont adds: “Audiences and fans love to focus on individual players or topics but it can be quite a long process today because you need to manually find the content. Very accurate video about one topic is quite easy to do with AI, because you can have a very fast deep dive on a gesture, or some action during the game, which was not really possible before. So that’s a way to boost fan engagement and to create gateways to key moments within the archive and monetise that with NFTs. We have one customer doing that – the National Rugby League in France. They are finding all the relevant pieces to transform content into NFTs to have this new vector of monetising.”
Room for improvement
There is, however, a general consensus that AI is still in its relative infancy and it shouldn’t be seen as the solution to all problems. “At the end of the day, it’s AI and AI’s not perfect, so we do put that caveat that it’s 75-80% and as we do more, as you process more, it will improve over a period of time,” highlights Krishna.
For that reason it’s important to be clear on why you’re using AI and the problem it will solve, rather than investing for the sake of it. “Do not focus on AI, focus on the goal. So what are you trying to achieve? We always say that AI is not a goal but a means to an end. The end can be more content, faster content, productivity; it can be a lot of different things,” explains Livadiotti.
Philippe Petitpont, Newsbridge
Importantly, no matter how good AI gets, there’s still a need for the human element, as Petitpont explains. “Lots of people are dreaming about some kind of AI that is able to understand everything in a game without any humans, but the technology is not ready yet to do that. When you are on the ground, you can see everything, you’re not limited to what people are filming. The limitation of AI is that even if you have robots on site that can look at everything, you will still be limited by the camera angle you have.
“Today AI is very good at looking at one camera angle, but very bad at looking at different angles because it’s different AI. So multimodal is key to being able to have that and to be able to have a good understanding of what’s happening in the game. That means building dedicated models for each sport and having a good vertical knowledge on that. So, to me, this is the next step, but not for tomorrow. Maybe for the day after tomorrow.”
Livadiotti adds: “Do not forget the human supervision. Look for a solution that leverages AI, but that allows for some level of human supervision to make sure that for high value sports properties, you’re not making mistakes. And, of course, on top of that, not only supervision, but the additional editorial piece, which is key in sports, because not everything can be automated. There is a lot happening in the game itself, but more and more when you look at the younger generations they are in some cases more interested by the shoulder content, what happens around the game, and this cannot be automated. AI shouldn’t be seen as a way to fire editors, it is a way to do the hard lifting, so that editors focus on what they do best, which is storytelling, which is the emotion around the game.”
Warech also believes the scope of AI will expand significantly. “AI will inevitably touch nearly every aspect of a sports broadcast and the entire sports ecosystem. As soon as content is captured, AI will automate the metadata tagging of this content to make it usable for multiple downstream applications. Without AI, property stakeholders cannot find and use content immediately for their own needs, nor can they easily offer it at scale for monetisation opportunities. That’s largely how it’s being used today, but its scope will continue to expand, mainly with content creation as Generative AI technologies and synthetic voice evolve to not only be discreet solutions but also complement one another, offering a new way to engage and reach fans through content.”
“AI shouldn’t be seen as a way to fire editors, it is a way to do the hard lifting, so that editors focus on what they do best, which is storytelling, which is the emotion around the game”
“I guess there are two areas which are, let’s say, about to happen,” adds Livadiotti. “One is personalisation, which is still not mainstream. It brings a lot of fan engagement, and therefore, of course, monetisation, but it’s not yet fully embraced by the actors in this domain. It is fully possible, meaning that the technology to build hyper personalised highlights is basically there, but it’s not yet a mainstream requirement for many different reasons. The other part is monetisation – the ability to make ads or sponsorship less intrusive is key. And then there’s ad insertion in live, detecting the less destructive moments in a given game to insert the ad, which today is easy to do manually, but it could be optimised with AI.
“Long-term evolution, of course, will be driven by Generative AI, which will be a revolution across the entire world. But across our own market, it already brings some interesting use cases that our customers are either asking about or that we’re investing in.”
For Magnifi, the focus now is on the breadth of its offering. Krishna reveals: “We want to spend a lot of time, at least in the short term, on really honing our AI development for the width of sports. Right now we cover about 30 but there are so many different sports. We’re working with new and upcoming sports because we want to make sure that we democratise editing for every sport and every fan.”
Warech concludes: ‘As we look ahead, Generative AI solutions like ChatGPT will drive increased inquisitiveness and foster exponential growth. When you add AR/VR technology to these AI ingredients, you get immersive experiences that the world has yet to see, but they are coming sooner than you think.”