Your browser is out-of-date!

Update your browser to view this website correctly. Update my browser now

×

AI’s future value is anyone’s guess… well, not really

While still viewed by many as an emerging technology, the reality of AI is the technology is being operationalised today at scale throughout the media and entertainment industry, writes TMT Insights' Brian Kenworthy

The media and entertainment (M&E) business is built on content libraries: Who has the most pristine library, and who can get it in front of audiences most efficiently while maintaining the highest quality? Our customers race for different ways to enable the globalisation of their content while searching for the best new technologies to support the distribution journey. Our industry thrives on forecasting which new workflow or production tools will stick around for the long term.

The current favourite is Artificial Intelligence (AI), and while many are still trying to estimate its long-term value, they only have to look at what’s happening today to see the benefits of AI in real-time.

The question isn’t whether or not AI will provide value, it’s how and where AI is providing value today. The answer is today’s technologists and innovators are leveraging this technology to provide efficiencies not only for current workload demands but to also get ahead of future production requirements in the near- and long-term future. 

M&E organisations are leveraging AI, and its subset machine learning (ML), for a range of business process functions, from executive forecasting to enabling project acceleration for day-to-day content operations. 

AI is already in widespread use throughout the industry, for various tasks such as automatic editorial detection, speech-to-text for caption creation, audio and video quality control as well as the regionalisation of scripts. 

A proof-of-concept in action

For tangible evidence, let’s point to a current success story involving a customer’s bulk catalogue delivery of more than 20,000 titles to a major global streaming platform, which was made more efficient through the use of AI/ML tools to perform speech-to-text, audio and subtitle conformance, and quality control (QC).

The objective was to prep their content library first for this platform specifically, and then also “normalise” it for the next distributor who might find an audience for the content. That entailed analysing the entire library from video to localised audio and subtitle components to uncover the typical patchwork of different frame rates, resolutions, and standards. Then, the task was to create some baseline of consistency for all titles to get the library to a place where it could be optimised, and monetised, for the maximum number of viewers. 

For example, a group of titles might be in an SRT subtitle format, which is not ideal for accommodating the maximum amount of distribution opportunities. It’s better to convert those to a more commonly used format such as DFXP or ITT. Also, we specifically used AI to analyse audio files to conform other audio files, in different languages, to the masters. We also leveraged this for subtitle conformance. 

Next, it’s all about avoiding duplicate efforts each time a library has to be prepped for a new fulfilment order. Now it’s done once and ready for repeat use, making the library format agnostic so it can be attributed to any content owner. And the best part is we can attach hard benefits to this AI-driven process in terms of time and cost savings. 

To complete this project manually, including the processing and conforming of at least 80,000 files, would have required the efforts of up to 10 different editors working over six months to a year. With AI, this initial phase was successfully completed and delivered in a matter of weeks.

Another customer recently leveraged AI for editorial use, analysing images, placing them on a timeline with other assets already cut for logos, bars, and tones. That allowed an operator to review and verify, make a few tweaks, and then ship it. They estimated a 50% reduction in the time needed to complete the project while freeing their current staff to focus on other, higher-value creative tasks.

Which leads to the elephant in the room: the lingering concerns of AI taking over people’s jobs. It’s important to note that in these cases, and in many others, AI/ML was aiding – not replacing – the creative human work by removing tedious and time-consuming tasks and in turn, enabling greater efficiencies and scale to support rapid content turnaround and delivery.

If we look back five or six years ago, there are many similarities to how people feared cloud technology and let their emotions hinder their adoption. After realising the transformation to new technology can be incremental versus a “light switch” occurrence, many of those fears have since been alleviated. 

With AI continuing to become more integrated into M&E workflows, people will warm up to the idea of AI assisting humans, especially once they experience first-hand how these tools enable efficiency and practicality, versus being a pie-in-the-sky science project.

This technology is not coming, it’s here now, with tools at our disposal to help drastically reduce project cycles from a year or more to something that can happen within a quarter.