Call for Speakers Sponsorships Contact BVE News

AI's impact on OTT services

2 January 2018

How will AI and machine learning impact OTT services and consumers in 2018?

Ahead of Streaming Forum 2018 (ExCeL London, 27 February), we asked a few of our expert speakers for their predictions on how AI and machine learning will impact OTT services in the coming months:

Head of YouTube Technology Solutions at Google, David Thorpe: I don't have any special insights into how they will impact OTT services, but I can certainly tell you how they should impact those services ... have an amazing selection of content available to viewers and ever-more sophisticated set-top boxes which add 4K, high frame rate sports content, maybe high definition content, etc. But the organisation of this content continues to be stuck in the 1990's with very little beyond a standard TV guide to inform both the providers and viewers on what's currently on, and what's coming next.

New techniques for analysing content within images and video are slowly coming online from cloud providers like Amazon and Google. In addition, research into graphics is coming to market with stabilization, face and object recognition is workable with the latest graphics processors. I feel we're finally going to get a step-change in how image and video content can be analysed, organised and presented to viewers, if the manufacturers of OTT devices can incorporate the latest graphics processing hardware into their devices.

I'm looking forward to the future where I can "follow" topics, people and events and my OTT box will record and playback content based on my interests, not just on what I "thumbs up".

CenturyLink’s Rory McVicar: Through 2018, I think we will see the influence of artificial intelligence grow – particularly within the world of live sports and real-time event broadcasts. With more and more avenues through which consumers can access content, we’re seeing a trend towards OTT providers seeking differentiation through overall experience – including interactivity. Machine-generated highlights reels based on consumer-defined themes, for example, could offer an additional layer of context for the consumer to weave into their own personal viewing experience.


Ryan Jespersen, WowzaMachine learning is about predicting the future based on the past. Providing a machine the ability to learn allows us to create understanding out of pattern recognition. There are a multitude of valuable use cases for Media and Broadcasters.

In certain countries, legal restrictions on inappropriate content can have serious economic consequences to broadcasters and OTT services. Creating AI software to watch for inappropriate content can trigger alarms to playout management in order to take specific actions, including:

  • Mark content as inappropriate through metadata
  • Blur video to remove the offending content
  • Edit or censure content

Currently, content moderation is a manual, expensive and cumbersome process for live video workflows and there are no processes in place for new media like user generated content. The ability to automate Metadata Enrichment can provide unprecedented value, including:

  • Speech to text and text to speech for accessibility use cases
  • Tagging tools, translation and text recognition
  • Recommendation engines
  • Automatic identification of faces, objects, logos, visual text to drive
  • Smart Advertising for product, ad and content placement    

Dom Robinson, Co-founder, id3as: While (contrary to popular thinking) the science of AI has not actually evolved in many great strides much over the past two decades, the data sources to which those sciences can be applied have scaled up beyond all recognition. The 'Big Data' explosion is now providing vast quantities of data in response to user activity and from this massive data set AI can observe trends and infer reactions that can mutually benefit the users and the service providers. Personalisation of service and better understanding of the target of a video (or audio) stream can help pave the way for better returns on investment from advertising (and to some extent subscription) models, and so naturally inferences driven by 'AI' as a key tool in 'Business Intelligence' can help service become more successful.

As processing of content itself also get faster so too will AI models be able to more quickly fingerprint content and map those fingerprints to rights management technology (watermarking etc), security systems (eg face, action, scene and product detection) and also network optimisation (traffic, compression, redundancy and so on).

Tommaso Cesano, Head of Business Strategy, Metaliquid: AI and machine learning will gain a primary role as their applications in OTT services are extremely powerful. Digital asset management and content discovery services will be the first to be massively innovated by AI.

Thanks to real time deep learning technologies, OTT players will introduce new functionalities that will change the way how we experience and interact with content and related information. AI will be a key factor in designing more engaging user experiences and revolutionize the way how we watch TV.

Erica Beavers, Head of Partnerships & Marketing, Streamroot: Machine learning has far-reaching implications for content delivery, from bitrate adjustments devices to per-device content sourcing. Early implementations have revealed much promise and we will continue to see the move towards highly customised delivery solutions in the coming year.

Meet these experts and others from Sky, BT, Amazon, BT Sport, The Football Association and more at Streaming Forum, 27 February, ExCeL London. Co-located with BVE Expo.


More News

Diamond Sponsor

Platinum Sponsors

Gold Sponsor

Panel Sponsor

Industry Partners

Media Partners

Conference Videos

View videos from previous Streaming Forum events.