When asked how major broadcasters might use generative AI in the coming months, most people conjure up dystopian visions Deepfakes of politicians To Minority Report-Style ads that follow you around the city. However, at this relatively early stage of the AI game, the reality is far less grim – and the biggest players are hoping it will prove far more valuable too.
Take, for example, the BBC’s live text pages: a popular if unglamorous service that converts current events, such as breaking news or sporting events, into real-time blogs. Automatically translating football match audio commentary into text isn’t exactly episode-worthy Black mirrorBut it’s one of the AI applications the world’s leading public broadcaster is most excited about as it deploys a dozen pilots across the company to explore the technology’s potential for the roughly 318 million people who tune in to it every week reached all over the world.
“[The audio to text project] “It opens up so many potential use cases for us to deliver value to our audiences through some very simple things like reformatting content,” explains Peter Archer, BBC program director for generative AI. “This is just really exciting for me. It may sound a bit prosaic, but if you can make this work at scale in a way that mitigates the risks but still respects the BBC’s editorial values, that will be a huge win.”
Freedom to experiment
Other possible successes being tested include a chatbot for internal knowledge management, tools for creating and translating headlines, and a public-facing BBC Assistant focused on educating children.
But according to Danielja Horak, head of AI research at BBC R&D, there is also “a much more ambitious story here.” Her team has been working on fine-tuning an open source model for large languages using proprietary data for a “very long time”, introducing increasingly sophisticated tasks as it adapts more closely to the BBC’s workflow, tone and style. “There [are] a range of things we do with AI to improve efficiency, save money, improve recommendations, transcribe things and search our archives; a whole suite of tools designed to reduce complexity so that people can actually do their jobs more effectively.”
Members of the editorial, business, legal and policy teams were also given individual freedom to explore how ChatGPT, Microsoft Copilot and Runway could add value to their roles. “This is a problem for the entire organization,” Archer says. “You have to let people experiment, you have to let people play. It’s so new, we’re all learning together, we’re elevating this with as much internal engagement as possible.”
Move calmly and don’t break anything
So far, so rosy. But as a publicly funded organization tasked with delivering “impartial, high quality and distinctive” output, the BBC is also aware of the need to keep staff enthusiasm within strict limits. Last October, the BBC’s country director, Rhodri Talfan Davies, set out a set of AI principles in which he stressed the importance of transparency, human creativity and social responsibility, while staff, freelancers and partners must adhere to strict editorial guidelines.
Archer sees this as both an opportunity and a limitation. In a situation where UK audiences spend more time on average per week per person watching BBC TV or iPlayer than Netflix, Disney+ and Amazon Prime combined, he believes the company has a unique opportunity to lead best practice in the broadcast space. “Nothing we do right now is automation without people,” he emphasizes. “The license fee is a privilege, so we have a responsibility for how we use new technologies. We absolutely have to be at the forefront.”
“Nothing we do right now is automation without people.”
Peter Archer, BBC program director for generative AI
After all, after the era of “move fast and break things,” which is certainly the case It broke just as much as it was repairedHowever, for commercial and moral reasons, surfing at the top of the innovation S-curve doesn’t always make sense. “We have what we call our Gen AI tech stack,” Archer explains. “We’re trying to think about what the architecture of that stack needs to look like, what specific capabilities in the organization we need, what LLM models we need, how we make sure we’re building in the most consistent way possible, and how we make that visible.” APIs for the rest of the organization.
“The opposite is that you suddenly realize you have nine different contracts with four different cloud providers, you can’t keep track of costs and someone has just issued a multi-million pound invoice for using ChatGPT. So we’re going to build it a certain way from the beginning and it may be a little slower, but that’s okay.”
Enemies under pressure
Another challenge lies in the dynamic between media organizations and technology companies, a special relationship as fraught as any geopolitical equivalent.
This month, Google launched trials of AI-generated search in the UK, and Pedro Cosa, a senior data expert who has led teams at Channel 4, WarnerMedia and News UK, sees the tension between discoverability and discretion as one of the most critical issues broadcasters face .
“The media companies have a very adversarial relationship with the technology platforms,” he says. “Nowadays people often check their news on Facebook, Google or Apple, and what comes up first is what makes you money or gets you clicks. So media companies really rely on these platforms to grow an audience, and the platforms rely on them to provide quality content, and AI is really making media companies think more about how they can balance that.”
But perhaps the biggest challenge facing the BBC – or any other organization – when it comes to AI is the extent of the unknowns.
Archer emphasizes that this is a time for leaders to question everything, remain humble and leave themselves open to surprises. “This is a long-term fundamental change in the media market,” he says. “Just like with the smartphone and the internet, we both need to look for the things that provide immediate value, but also the things that expand us and the industry in terms of our understanding of where this technology could go.”
Of course, for broadcasters who can’t hold license payers accountable, AI could certainly go places that Auntie might shy away from.
Personalized TV shows that adapt to your viewing history and preferences already exist, while LA-based startup Channel 1 plans to stream AI-generated news presented by multilingual bots by the end of the year. However, for football fans around the world, these jazzed up text pages could well have a bigger impact on their daily lives. And no one will lose their job because of it.