BBC unveils blueprint for long-term research driven by AI and IP
The BBC has outlined how it will plan future R&D and it is being built around internet-only, intelligent, intermediated, interactive and immersive media.
“These are the forces shaping the media landscape for the next generation,” says Jatin Aythora, Director, BBC Research & Development.
Framing its focus and investment going forward are three hypotheses, one of which states that ‘all user interaction will be driven by artificial intelligence’.
The BBC itself is in transition, along with other UK broadcasters, from over the air terrestrial distribution to one that increasingly relies on broadband networks.
“Our work at R&D has become about more than just preparing the BBC for that transition,” says Aythora. “In our new research cycle, we need to explore the use of AI more fully to ensure that how the BBC uses this technology is always responsible, ethical and in the public interest.”
Funded by the British public in an annual licence fee (currently £174.50 / U$231 per household) everything the BBC and by extension its R&D wing does must meet its public service remit.
As Aythora puts it, the work of R&D must be “ethical and for the good of the BBC, licence fee payers, the media industry, and society” – that’s quite a responsibility but one BBC R&D has turned to its advantage, benefiting the wider industry too.
It beat most of the world to the streaming age when iPlayer launched in 2007. Before that it developed one of the first video streaming codecs and its computer vision work is the basis of ubiquitous sports graphics system Piero. More than a decade ago it pioneered interactive and personalised media. BBC R&D was one of the main contributors to industry work in DVB updating the DVB DASH standard to support consistent low latency streaming.
More recently, it helped establish key sets of principals and tools to combat AI misinformation with the Coalition for Content Provenance and Authenticity (C2PA).
So how it thinks the technology landscape will develop is worth listening to.
Three hypotheses are central to its current thinking:
All user interaction will be driven by artificial intelligence.
From content recommendations to real-time language translation and automated story generation, AI will become central to shaping the user experience. BBC R&D thinks AI-driven interaction will allow for “hyper-personalised content delivery, improved user retention, and optimised marketing strategies” but this also presents a challenge for the BBC.
“Business models are changing and pushing industries to transform,” says Aythora. “As the AI ecosystem evolves, we will need to define an approach that ensures we can maintain and build on the trust we have in this new environment.”
-ORG.png)
Immersive gaming concert experience: Avatars could watch views of the orchestra and conductor throughout different zones of the virtual venue including the interactive galleries (left), the courtyard (top right), and play zone (lower right).
The internet backbone will feature a new layer that provides a permanent and transparent registry for digital assets.
This hypothesis envisions using distributed ledger technologies (such as those that underpin blockchains), to create a new layer of core internet infrastructure that provides secure and tamper-proof tracking of digital content and assets. Such a layer can help fight disinformation by verifying content authenticity while ensuring appropriate attribution and rights management for creators.
“Tackling misinformation, content piracy, and copyright infringements are key challenges for the For the news and media industries,” he says. “This new layer of internet infrastructure would promote greater transparency and trust, empowering both creators and consumers while fostering a more credible media ecosystem.”
AI infrastructure will be treated as national critical infrastructure.
The third working assumption is that AI’s underlying infrastructure--data centres, training models, and compute power--will be recognized as vital to national security and economic stability, much like how utilities such as electricity or telecommunications are managed.
“For the media industry, robust AI infrastructure will ensure the seamless operation of tools used to generate, curate, and distribute content,” says Aythora.
It would also guarantee resilience against cyber-attacks targeting AI systems and uphold the reliability of AI-driven journalism during wars or elections.
From these hypotheses, BBC R&D then worked out how they might impact the BBC, how they relate to its public service remit and how the BBC could contribute to any associated tech development.
“We don’t assume that any single technology will define the media ecosystem of the future,” says the BBC R&D chief. “Instead, we believe that it will reflect a set of outcomes enabled by different types of technologies.”
Its activity is guided by a vision of a media landscape that is increasingly Internet-only; Intelligent; Intermediated; Interactive and Immersive. These five ‘I’s (not the five eyes of the Angle-Australian-UK spy network) are outlined as follows:
Internet-only - Produced and distributed by IP
When over half of UK households are predicted to exclusively watch TV online by 2030 and it’s almost a matter of policy that BBC itself will evolve into a primarily internet-based organisation over time, it’s stands to reason that the next generation of audiences will consume BBC content over IP.
BBC R&D is exploring how “to ensure that the future of media provision is sustainable, affordable and universal and that the organisation is positioned to create and distribute content in ways audiences want.”
Examples in sustainability include a tool called LECCIE that allows BBC teams to monitor and minimize their cloud energy footprints.
It claims to have improved streaming by making open source contributions to the dash.js JavaScript DASH player to improve low latency support as well as testing and assessing different stream switching algorithms through simulations and trials.
New internet experiences include a music discovery service, designed by BBC R&D “to bring back some of fun and excitement of flipping through records in a bargain bin when discovering new music. It’s all well and good to take a recommendation from the record shop, but you don’t really know if you like it until you listen and the music connects with you, sometimes in unexpected ways.”
The BBC music discovery pilot, Orbit
Intelligent – Enhanced and automated by AI
BBC R&D is investigating how Generative AI in particular can be used within the organisation with the provisos that its use is always in the best interests of the public; always prioritises talent and creativity; and is open and transparent with audiences when used in content creation.
“We want to develop AI technology that has been trained ethically and responsibly and will enhance the output that our journalists and producers create,” says Aythora.
Examples here are the trial of a speech-to-text GenAI tool called Whisper AI to quickly generate a high-quality transcript of audio. This is reviewed by a member of the editorial team, and edited where necessary. A final transcript is then uploaded with the audio online.
The output of speech to text tool Whisper AI
Intermediated – Mediated by platforms and agents
According to media regulator Ofcom over half of UK adults use social media as a news source. Therefore, as the nation’s most trusted news source, the BBC desires to maintain a direct, independent relationship with audiences. “The BBC needs to understand and work with the platforms and agents that sit between us and them.”
Examples here include Freely, the free streaming service that launched a year ago and offers live and on-demand TV. The BBC worked with other UK broadcasters to deliver this and contributed to the development of the HBBTV operating system for smart TVs.
BBC R&D also played a foundational role in the 'content credentials' C2PA initiative to tackle misinformation online, accelerated by AI. Other members include Adobe and Microsoft. C2PA has developed a technical standard to encode information about the provenance of content that show where a piece of media has come from, and how it’s been edited.
Interactive - Optimized for user experience & engagement
Acknowledging that audiences demand more interactive experiences, R&D has been developing technologies that enhance audience interactivity while supporting the UK’s creator economy.
A key project is Wing Watch. This interactive wildlife stream uses ML-driven data to help audiences navigate live content based on their interests. “This type of personalized interactivity enhances user experience, increases content discoverability, and lays the foundation for future AI-driven engagement models,” says Aythora.
Wing Watch shows viewers extra data such as AI-automated jump points on the timeline for common bird species.
Immersive - Augmented by immersion & contextual awareness
Along with enhanced interaction, BBC R&D is exploring how it can create new audience experiences that are immersive, for example using data about contextual awareness sent from the user’s connected device. Many of its experiments in this area are around live music, such as bbc bringing live concerts into multi-player online game experiences.
One strand of this work involves investigating the use of volumetric capture technology. The BBC say this “ideal for intimate performances of one or two artists under controlled lighting conditions” and allows fans to view the performance remotely.
It is also researching approaches using broadcast cameras as a part of work in the MAX-R EU collaborative project. “This approach may be better suited for a larger number of performers, with complex stage and lighting conditions, at the expense of a less 3D feel and a more limited range of viewpoints,” it says.
Another augmented reality prototype uses spatial audio, geolocation, voice recognition and physical gestures to take a user wearing smart headphones “on a journey through universes without ever leaving your local park.”
Taken together and the hypotheses and 5is are the blueprint for the BBC’s long-term research agenda.
Aythora says, “By focusing on AI-driven user experiences, ensuring transparency in digital asset management, and recognising the strategic importance of AI infrastructure, media organisations can adapt to a rapidly evolving technological and societal landscape while maintaining relevance and trust.”
Related Articles
Jake Ward, Business Development Director at Groovy Gecko, breaks down the current state of the media & entertainment industry in 2024, including Netflix's 'What We Watched' report, Amazon's ascendancy, Apple TV+'s value as a value-add, the BBC iPlayer's dominance, and more.
02 Apr 2024