Comment


Entertainment marketers: AI won’t steal your jobs, but get ready for change

Amid a popular culture rife with chilling, dystopian portrayals of artificial intelligence, you may be scared spineless about the technology’s impact on your relevance in the workplace, and who could blame you? Yet, there’s evidence that AI is actually creating new jobs, with a recent study estimating 2.3 million AI-generated jobs by 2020, far more than will go away.

What does this mean for entertainment marketers? The odds are, you’ll keep your job, but find it to be very different. The question is just how that will happen and – more important – how should you prepare for it.

Having built and managed numerous OTT platforms over the years, I predict that 90% of all visual marketing assets for feature films will be created by AI in the next two years. Not only that, but it will be deployed for taking content in old formats and transitioning it into new versions for localisation and things like compliance editing, where the system uses AI to build versions of movies that will be compliant in different countries.

Much like Apple uses facial recognition to open your phone or Google uses item recognition to help you identify a type of flower or painting you’re observing, AI will be able to analyse a feature film (or television series/special) and generate still and video marketing elements tailored to a campaign’s goals.

In the cases of Apple and Google, AI has been “trained” to capture certain sets of information and create definition through it. For a feature film, it’s about training AI to build rich metadata around each scene and even each frame. For example, an explosion gets attached with a particular piece of metadata, which is then categorised as “action.” If AI captures 10 scenes it identifies as “action,” it then categorises the movie as an action movie.

This technology is already at play with Netflix, for which its main relevance is customer retention. Netflix uses AI to build a user experience that’s tailored to each person. And while, two users may have the same set of shows suggested to them, the imagery could differ greatly depending on their viewing habits.

And this type of AI implementation helps with not only the discovery process – auto-categorising content and displaying it accordingly – but also with marketing.

Using AI to generate visual marketing assets means marketers will spend considerably less time focused on the monotonous tasks of searching, sorting and pulling assets – time they can devote to a higher level of strategy. Done manually, the asset pulling process takes two to three hours for the typical feature film, and while a professional could skip ahead in the movie looking for scenes, they still have to slow down and then manually pull the scenes they want.

Surprisingly, an AI-generated system for pulling visual marketing assets will also give marketers more control of the process. They can leverage scalable analysis into more effective, targeted campaigns capturing consumers and merchandising products.

It means hyper-targeting will become common and open the possibility of creating marketing assets that enable the brand to reach more niche sub-audiences. For example, the AI for sci-fi movies will be able to pull video segments and images targeted at sci-fi fans (image of alien), at romance fans (image of titular couple kissing), action fans (battle images) and dark comedy fans (image of villain making a joke at the expense of the hero).

Further, AI won’t merely provide entertainment marketers the assets they plan to use, but it will also give them the metadata associated with those assets, which can then be used for targeting parameters on digital ads. Hypothetically, if a marketer receives four different assets – two for action fans and two for romance fans — the metadata for the first action scene could include additional descriptors such as “martial arts” and “Hong Kong.” These parameters could then be plugged within a Facebook ad audience segment: action movie lovers – Hong Kong martial arts. Such steps would greatly increase the effectiveness of their campaign while reducing the amount of manual work required.

With the average major studio fielding hundreds of movies annually, this technology will enable marketers to triage between those films requiring the most attention and those less demanding of investment. For the highest priorities, marketers could still apply a more hands-on approach and manually choose the absolute best segments. But, for the rest, they rely on the AI do most of the work.

Don’t fear AI, but anticipate the changes and adjust strategies to best capitalise on this future. Ideally, this means working closely with sales and production teams to triage the films as accurately as possible and also learning how AI works to help train its algorithms.

We’re very excited by the technology and working towards making it a reality. Are you ready?

Aaron Sloman is chief technology officer, OWNZONES Media Network.