Artificial Intelligence is not a production panacea

Artificial Intelligence (AI) and automation carry huge potential benefits for the video industry. There are numerous production scenarios in which machine learning and automation can enhance existing processes for the creation, management and distribution of content – however, introducing AI to video production needs careful planning.

The contextual power of content should not be lost in the pursuit of efficiency. We can already see the benefits in transforming certain aspects of production, such as the introduction of machine learning into video encoding, which has removed the limitations of hardware-based systems and allows video services to be delivered in high fidelity across the globe.

Machine learning is also useful for content creation. Video editors are able to train automated algorithms to recognize certain ‘triggers’ within recorded or live video inputs, and instantly create short clips, sequences or apply overlays and effects without human intervention. We’ve also seen how computer vision can help editors to quickly understand the context of videos and images, and cognitive audio services can process speech-to-text capture and translations, speeding up the process for creating and distributing content.

While these applications of AI and machine learning stand to revolutionise content workflows, it is increasingly clear that human insight is as important, if not more important, in this digital transformation. Simply put, computers aren’t great storytellers.

Take sports broadcasts as an example; for football, data inputs can tell you when a team scores, or when a foul is committed, but it can’t recognise a piece of dazzling skill or the best reaction shots from the crowd.

Audio analysis can tell you the crowd is cheering, but it unlikely to differentiate between a goal celebration and a pitch invasion. Understanding the context of your content matters.

There is real value in automating production for the key moments, such as goals or great saves in this football example, which helps publishers get to market first with content. But it’s harder to create an engaging story of the game without using other contextual references, which helps viewers connect to the content.

The same goes for drama, action, or almost any genre of content. A computer can recognise your lead character’s face and tell you when it’s in-frame, perhaps apply effects or music, but it can’t read the context of the scene they are in, which means it also can’t recognise the best moments of dialogue or the most heartfelt scenes, in order to react appropriately.

Human editors, equipped with the tools to quickly and easily access music, layers, clips or graphics associated with a certain character or event, are able to create a more impactful piece of content, spending less time on menial tasks and more time focused on their art. AI and machine-learning can expedite production, but they don’t remove all the people from the process.

We must view AI in this frame – as a gradual empowerment tool. Its development should be shaped around humans.

In today’s multi-platform market, the competition for viewers is ruthless. The challenges that video editors face have changed – content must be highly tailored for multiple audiences and optimised for multiple platforms. For news, sport or live events, speed of delivery needs to be reduced significantly, demand for content is immediate and pervasive – ‘real time’ clips are those that win the biggest audiences across social media and mobile platforms.

All of these challenges must be met while creating unique, meaningful content. AI can help solve these problems, but it must do so in a way that leaves human engagement at the centre of production.

Despite the significant improvements in video-based AI, it will be some time before AI can comprehend the complex subtleties of emotion and human interaction in every language, system or social context. Until then, AI is a powerful tool to make video production faster, easier and more efficient, but as a partner…not a panacea.

Gareth Capon is CEO of digital video production specialist/platform Grabyo