AI isn’t a far off dream—it’s a present reality. Machine learning algorithms and neural networks now have their wiry digital hands in everything from self-driving cars to detecting credit card fraud and determining what we watch.
Why does AI determine what we watch? Because there’s just such an incredible amount of video out there. Video now accounts for 74 percent of all internet traffic and that will rise to 82 percent by 2021 according to Cisco. And while almost half a million minutes of video are uploaded to YouTube alone every day, not all of it is worthwhile (we rest our case on PewDiePie).
Consumers’ attention spans are maxed out. That’s why AI is being applied to both create video content and help users find more of what they’re looking for. But is it good for us?
AI Helps Write the Script
Imagine being a Hollywood producer pressed to find the next big blockbuster. How could you predict what audiences might want? Increasingly, studios are turning to AI for answers. According to The New York Times, the analytics firm Epagogix offers a neural network that ingests scripts and evaluates them based on plot points and historical box-office data. It does more than give a critic’s thumbs up or down: it can provide specific recommendations to increase filmmaking ROI. And this is only the beginning.
Video for both consumers and for business is being gently guided by AI recommendations. Netflix used machine learning algorithms to determine the plot and actors of the original hit series House of Cards, reports The New York Times. On a more commercial scale, marketers are using smart analytics to determine what video elements are most likely to entertain viewers and go viral. Vidyard’s video analytics, for example, can peer into user behavior to see how much they watch, what parts they watched, and what segments they revisited. Marketers armed with these automated insights can iteratively improve their videos for maximum consumption.
AI Serves up the Videos
The videos you watch are also increasingly selected for you by AI. Marketers have long experimented with using algorithms to deliver personalized experiences via the web and to deliver ads. This same technology is increasingly being used to interpret and serve up video.
AI armed with computer vision and natural language processing is now savvy enough to tell what’s happening during videos. This wasn’t the case until recently and videos were something of a black-box when it came to metadata. Marketers had to tag videos the same way they did images and tagging often couldn’t capture even a fraction of the detail. Nor does tagging and curation scale well for large enterprises with thousands of videos. But curation by AI does.
Microsoft’s new Story Remix app, soon to be released on Windows 10, can recognize objects well enough to apply animations over them, according to CNBC. Facebook’s AI is smart enough to tag your friends in videos and marketing technology startups like Clarifai, Valossa, and IRIS.TV all use AI to provide rich tags for videos which identify objects, sounds, people, locations, and all variety of context. By marrying user data with rich video data, brands can offer hyper-personalized video recommendations.
Publisher sites everywhere use this AI to serve up videos. Facebook, YouTube, and Netflix famously do it, but now so do news sites, online publications, digital magazines, and virtually any media company with a reasonably sized repository of video. IRIS.TV clients like Time Inc, Hollywood Reporter, and Billboard have seen an average 70 percent increase in views after only a few months.
The Implications of AI for Video
More and more, the videos you watch are both written and selected by AI, and all fears of killer robots aside, marketers will find that this is both good for consumers and brands. Marketers can deliver lower volume but higher value content because they’re no longer having to throw spaghetti at a wall and consumers are finding less disruption when video recommendations are increasingly relevant. Are the videos you watch dictated by AI? Absolutely, and for the time being that seems to be a pretty good thing.