AI and the evolution of MAM: Part three – future outlook

The outlook for AI appears as complex as ever. Where it concerns the authenticity of content and metadata, AI is at work on both sides of the ethical fence; playing miscreant and detective. For all of its capabilities as a democratised technology in the hands of everyday folk, and certainly in the wrong sorts of hands where it concerns deepfakes, its professional applications within media asset management (MAM) systems seem to point to a future where it can play a significant role in the tracking, sourcing, and authenticating of content.

One small step for MAM…

With that in mind, can AI be a giant leap for MAM-kind? What newfound capabilities are we likely to see in the coming years as the technology becomes more fully adopted?

Jonathan Lunness, Blue Lucy’s Chief Strategy Officer, points to two key areas: new revenue streams and compliance. “Within a MAM system, AI could be used to help content owners monetise their archive by identifying library content that ties into current events or trends. For example, if a celebrity is trending on social media, the system could automatically surface any content featuring the celebrity or clips where they are mentioned.

“In the same way that MAM systems are often integrated with rights management tools to ensure that licensing agreements are adhered to, they could also become instrumental in tracking where content has been created using AI tools to comply with different distributors’ policies regarding the use of AI content,” he suggests. “Ironically, AI tools might even be instrumental in opening up this new capability to identify AI-generated sources and track the temporal metadata to identify the exact points in affected clips.”

AI-assisted search, content discovery, and recommendation engines will address many of the challenges users face with today’s MAM systems – according to Ben Davenport, Founder of Beyond the Grapevine – aligning them with the user experiences we’ve come to expect from web search engines.

“Equally important, though perhaps less ‘shiny’, is AI’s potential to optimise workflows and streamline the media supply chain,” he says. “By analysing and identifying trends in content usage – and even predicting future needs – AI can help find efficiencies in content transformation and transfer, such as transcoding or moving content between storage tiers. This could reduce costs for media organisations while still supporting fast-track or ‘red line’ workflows for high-priority content.”

For Media Asset Capital’s Jeremy Bancroft, AI content generation has reached an astonishing level. “The creation of new content from a series of prompts still leaves me in awe. One of the most useful applications of this is the ability to extend the head and tails of shots, thereby providing the editor with greater choice. Future capabilities may well include alternative, virtual camera positions, and changes in lighting effects.

“It is unclear to me where these capabilities will lie,” he muses, “in the MAM system or in the editor: most likely both. For news and current affairs programmes, the use of AI-generated content is problematic and could undermine the reputation of a news organisation. In these applications, the ability to quickly find archived content that is relevant, and that will match its style to a current story will be game-changing.”

There is also the belief that integration with multiple AI capabilities will eventually, and massively, drive down localisation costs. According to technology and operations consultant Rod Fairweather: “When we can regenerate original scripts, and recreate them in foreign languages with matching emotion of delivery, combined with lip syncing and reusable character voices, and achieve this in a very short timescale, our higher value assets libraries become more monetisable.”

Content provenance

“Future AI will play a part in determining the chain of trust and authenticity of content. I don’t believe that humans have the time or the skill to carry out this kind of validation.” Bruce Devlin, Mr MXF

Fairweather also points to blockchain as a potential enabling force, and that while its role in media “may sometimes feel like a solution looking for a problem, it has potential to secure IP rights, and confirm content originality,” he suggests. “In a world that is getting flooded with AI-generated content, authenticity may become one of the key attributes to any footage.”

With trust in the veracity of content currently plaguing news organisations the world over, it seems that we are already at the point where authenticity is key, if not vital, especially for footage obtained outside of the traditional affiliate networks and trusted agencies such as Associated Press and Reuters.

ITN’s Tami Hoffman previously told IBC365: “We want our journalists to continue using traditional journalistic practices. There is a role for technology… but technology is not going to be the sole solution.”

It does seem, however, that while it may not be a silver bullet, its impact is still likely to be compelling. “Future AI will play a part in determining the chain of trust and authenticity of content,” states Bruce Devlin, Founder of Mr MXF and SMPTE Fellow. “I don’t believe that humans have the time or the skill to carry out this kind of validation, and a mix of AI correlation or sleuthing with good, clean metadata practice should be able to reveal the audit trail of any frame of content.”

The use of AI to better establish truth and authenticity in news is a controversial area. “Public service broadcasters especially face an increasing obligation to check and cross-check their many sources of information and content before they can create a secure and compelling analysis editorial,” suggests Russell Grute, Managing Partner at Broadcast Innovation.

“AI services can help with the scale and depth required and yet AI services are also part of the problem. AI is already widely abused in both social media and increasingly anti-social media to create alternative narratives.”

Alternative protagonists

There’s no doubt that AI’s influence on the MAM environment is both currently significant and potentially game-changing. But what other driving forces will be at play in the evolution of this core component of the media workflow?

“I’ve been thinking for some time that the technologies enabling spatial audio and immersive video are very much the Cinderellas at this year’s AI ball,” ponders Grute. “In live sports, spatial audio is adding a lot to improve media workflow and presentation of the commentary, crowd sound and the sports action. Salsa Sound and improvements in Dante integrations have been catching my ears.”

Meanwhile, he says, viewers are finally impressed with the next generation of immersive and interactive services. “What’s changed is that vendors including Apple and Dolby can put together the whole value chain to produce and distribute immersive spatial media. In addition, dramatically improved graphics hardware and headsets are now available cost-effectively through convergence with gaming.”

Davenport pits his perspective against a backdrop of media organisations needing to undergo rapid transformation to stay competitive with video consumption trends moving towards the likes of TikTok and YouTube.

“I’m going to be controversial and suggest the real challenge lies in an outdated approach to technology adoption,” he says. “Many media companies have treated MAM and other critical systems like traditional hardware investments; large, overarching contracts with 5–10-year replacement cycles. This approach simply doesn’t match the fast-paced, ever-evolving demands of today’s media landscape,” he says.

“One of the most promising technologies to address this is the use of enterprise service bus (ESB)-based, low or no-code media integration platforms. By separating MAM and other front-end applications from the underlying infrastructure, these platforms enable media organisations to implement new software solutions in smaller, more targeted projects.

“This modular approach reduces reliance on a single vendor and minimises the risk of vendor lock-in. More importantly, it allows organisations to respond to changing market forces more swiftly and even proactively.”

Truth and consolidation

“If one takes cloud deployment as a given, and AI as emerging, I think that the bigger change is going to be in the number of suppliers offering MAM solutions.” Jeremy Bancroft, Media Asset Capital

 A note of caution raised more than once in this feature series has touched on the prospect of AI feeding off datasets infiltrated with AI-generated content or metadata; effectively, AI feeding off itself. This is something that Devlin describes as “beyond dumb” in many circumstances.

“Fundamentally, AI needs lots of high-quality metadata to produce valid outputs, free of the worst erroneous correlations of seemingly matching data (i.e. hallucinations). We need to develop media-metadata workflows that deliver large volumes of certified truth into data lakes for AIs to feed on. That data can be everything from spatial positions of actors on stages, lighting, weather, dialogue, audience figures, etc.”

“AI is a useful tool,” he continues. “AI working on cheap, eco-friendly, low power authenticated data sources will always beat AI feeding off its own outputs.”

According to Lunness, the ongoing proliferation of distribution platforms is a big factor in the evolution of MAM. “It has resulted in an increase in both the amount of content and the number of versions of each piece of content that now needs to be managed within systems like BLAM,” he says. “Changing viewing habits and the move to vertical and shortform content have exacerbated this situation and it’s never been more important to be able to track all the different versions of an asset and find the original. After all, you can’t monetise what you can’t see!”

Bancroft’s final observation takes us towards a consolidated supplier landscape. He concludes: “If one takes cloud deployment as a given, and AI as emerging, I think that the bigger change is going to be in the number of suppliers offering MAM solutions.

“There are simply too many providers offering similar solutions. There are outliers, such as Dalet, who can offer highly customised platforms, but the majority of the other vendors are difficult to differentiate between. I think that we will see significant consolidation in this space.”