In your opinion, how is AI helping to reshape the way video and audio are produced?
Content is now needed well beyond the traditional single stream—we’ve seen translations, regional accents delivered, subtitle generation and other quality and accessibility tools at a speed and pace that has fundamentally changed who content can reach and how.
AI has reduced staffing and infrastructure constraints. In sport, we are now seeing the ability for smaller organisations to run automated camera coverage with minimal crew. As the demand for content increases, the agility of these smaller organisations, unrestricted by legacy hardware or operational mindset, become a competitive advantage. The ceiling for quality is within reach now, and that changes who actually gets to tell the stories. From a perspective of inclusion and representation that’s important to me and a fairly underappreciated impact of AI right now.
REGISTER NOW FOR THE AV AND BROADCAST SUMMIT
What are the specific areas where AI can have a major impact? (live, post production, subtitles, PTZ cameras etc.)
For me, the starting point is the audience. In live we already see the impact of live subtitling and alternative commentary as a real win for underserved audiences and accessibility, not just technological advancements.
Post production crafts are not being replaced; repetitive tasks that are time-consuming and not creative are being automated. The interesting impact is around the integration of processes which were previously done in isolation. Reductions in repetition and delays caused by manual sign-offs mean content can get to air quicker at lower operational cost.
Archive logging/retrieval/preservation holds a vital place culturally for preserving the history, arts, language of all humans. Making these discoverable and retrievable through AI is historically significant. We’ve learnt from the training models used and the lack of diverse training material needed to avoid hallucinations and bias.
What are the key risks around trust, accuracy and copyright—and how do you validate AI outputs in professional or mission-critical environments?
It’s about trust, copyright, data provenance and licensing. Content such as news, factual, and live will still require human verification built into that workflow.
We’ve seen C2PA become the industry standard for embedding verifiable credentials into content at the point of creation. Fingerprinting and watermarking have become established methods of content verification and a number of news organisations have their own Verify tool, whether it’s totally AI-driven or has some human oversight within that process. AI generates, humans decide, and provenance is documented at every stage. If you build governance around that chain, then you’ve addressed all three risks.
How does AI change the underlying technology stack—compute, cloud vs edge, networking —and what does that mean for system design going forward?
There’s no right or wrong architecture anymore.
I think of it as: public gives you elasticity, on-prem gives you consistency and edge provides immediacy. So the skill is knowing which works where for your production. In practice, that means edge for time-critical, location-based immersive experiences. On-prem for sensitive data, sovereignty. Cloud for major events on demand. Having this flexible, modular approach allows you to increase requirements and scale on demand whilst retaining control of sensitive data locally.
The network layer is where we will see redesign, becoming an adaptive and active part of the production pipeline. For example sport coverage with AI camera tracking, analysis, and highlights creation. All at once we’ll need to prioritise as these spike and shift. Looking further ahead, 6G is being designed as AI native from the ground up, network resources, spectrum allocation and processing power will be allocated in real time based on AI predictions.
REGISTER NOW FOR THE AV AND BROADCAST SUMMIT
Could AI lead to a skills drain in both broadcast and pro AV?
In a time where the industry is converging with pro AV as well as an increase in personalised content which serves a multitude of platforms, I don’t fear a skills drain, just a reset of skills and opportunity. This reshaping of workflows across both is a catalyst for opportunity. We have seen this major change across our sector many times, analogue to digital, on-prem to cloud. The experience around the introduction of 2110 and NDI opened those conversations about how we reskill people from IT engineering backgrounds to understand broadcast, so it’s about how we have learnt from these changes and apply those learnings to this. AI can automate a task but can’t replace the human who understands the why (yet!).
Young people still want to work within the sector and industry initiatives like On Air last year were a huge showcase of the international appetite for careers within our sector, and its success means the plans this year are even bigger.
Through my engagement with The Baker Dearing educational trust, I see an industry-led approach to skills and preparedness for industry. New entrants to our sector need not only the technical proficiencies but also human skills like critical judgment, being able to navigate change and adaptability.
I’d describe it as ‘curiosity over specialism’.
How do you think fears around the growth of AI should be addressed?
Leaders need to understand those fears and recognise what teams might be feeling. Preparedness is key, 94 per cent of broadcasters want to enable AI, but only 15 per cent are ready, according to a Newscast survey in January of this year. Deloitte’s State of AI in Enterprise report shows a lack of AI skills as one of the biggest barriers. As leaders, we need to invest in people as well as technology, addressing the gap between expectation and readiness, as that’s where the anxiety rests.
Honesty and clarity around where the use of AI is likely to impact roles and the why; some roles will change significantly and some will disappear. Governance within organisations is non-negotiable, particularly in autonomous AI systems.
Data sovereignty should be considered, international workflows, who controls the training data used and ultimately any data/IP. Is the infrastructure you’re building truly yours, stable and safe?
Ultimately, human oversight is a competitive advantage in workflows. It creates trust for teams and audiences.
