If you’re impressed by today’s “smart TVs” that handle all of your digital entertainment needs, just wait for the next generation of TVs powered by artificial intelligence (AI). In fact, the Chinese tech company Xiaomi recently released the “first AI television” (the Mi TV 3S), and seemingly every month, a new company is highlighting how artificial intelligence could be used to transform the TV viewing experience. So what’s the future of artificial intelligence and TV?
The first and most obvious way that AI is going to change the TV viewing experience is via advertising. Ads are going to become super-targeted, thanks to the ability of powerful AI algorithms to deliver the right ad, at the right time, to the right person.
Think about the current viewing experience on linear TV – each viewer sees the same ad, regardless of geographic location, how much money they make, or their age, race, gender or ethnic background. Thus, during prime time TV, tens of millions of people are all seeing the same ad at the same time.
Now contrast that to the digital streaming experience, such as you might experience on YouTube. Here, it’s possible to take information that YouTube already knows about you to show you specific ads during the pre-roll and post-roll. And, since YouTube and Google search are both part of Alphabet (the company formerly known as Google), it’s possible to take some Google search data to show you an even better assortment of ads. For example, if you recently searched for “Lexus” on Google, then you might very well get served up a Lexus car ad on your next YouTube clip.
With AI, it’s possible to make even finer-grained inferences about you, based on your entire digital entertainment viewing history and which device you’re using to stream TV. There’s more data to crunch, and more information to put into a complex algorithm. The ultimate result? A TV viewing experience where ads are adjusted on the fly, based on time of day, geography and possibly even your mood.
Another area where AI could make inroads involves the future of TV content. Think about the way traditional movies are made – for every hour of content you see in the cinema, there’s plenty more content that’s left on the cutting room floor. In short, human editors decide what to include and what not to include. But who says that humans are best at this? It could be the case that an AI-powered algorithm might be able to assemble a better film or TV show based on content that’s been filmed already.
As a proof of concept, 20th Century Fox teamed up with IBM Watson to create the first-ever AI-powered trailer for an AI-themed film, “Morgan.” Admittedly, it was rather gimmicky – but the idea is an intriguing one: an AI-powered algorithm might be uniquely able to figure out which scenes and which lines of dialogue should be combined into a trailer to create the biggest buzz and anticipation about the film.
In many ways, it’s no different than letting AI-powered algorithms make slideshows and short videos using your photos. Often, these slideshows are quite good, with actions in each photo matching up quite nicely with musical clips. Take that to a logical extreme, and you have the future of AI-powered content.
If you think about the way most people decide what to watch on TV, it’s still fairly simple – either they flip through all 100 channels as fast as they can, hoping to find something that sparks their interest, or they consult a regular TV listings guide in a nice, tidy grid format. But with new AI algorithms, it’s getting increasingly possible to predict what you want to watch, and show it to you, without you even having to choose.
In some cases, some satellite TV companies – such as DIRECTV – are working on technology that will automatically record shows for you on your DVR, all based on your past viewing preferences. The idea is simple: there are so many shows out there, that it would be nice to have a really smart companion (i.e. your AI-powered DVR) keep track of the best shows and record them for you.
And think about the Netflix recommendation algorithm. Thanks to machine learning algorithms, Netflix is becoming very good at predicting what you want to watch. There were some hiccups at the beginning – especially if several people shared the same Netflix account – but Netflix is getting better and better at figuring out your personalized viewing habits
In one creative use of AI technology, CSPAN is using Amazon’s voice recognition system to identify and mark up all content appearing on the network. An AI-powered bot is literally able to label all the content in such a way that any snippet of speech or reference can be found very easily. Nobody really wants to sit through hours of Senate hearings, right?
So it’s easy to see how this is going to be a real time-saver. Imagine, for example, you’re working on a report for a major think tank in Washington, and need to find all the times that President Trump has ever referred to “healthcare” and “Hillary Clinton” in the same speech – you would have a very easy way to skip to the good stuff, rather than wade through hours of speeches, testimony, or events. In short, CSPAN is creating a very powerful searchable video library, and it’s all made possible by AI.
You may not realize it, but your TV is already hooked up to the Internet of Things, at least indirectly, in terms of objects like your set-top box or DVR. And that trend is only going to accelerate, as more companies look for ways to enter your home via the TV.
One great example is the use of voice search and voice commands to control the TV viewing experience. Every voice command can only be analyzed, thanks to sophisticated AI algorithms. A computer has to interpret the sound you make with your voice, transform those words into digital signals, analyze those signals, and then respond, all in near real-time. Thus, you can already give your TV certain commands like “Rewind 30 seconds” – and the TV will follow your instructions. You won’t have to press any buttons on a remote control, everything is handled via a voice command.
Or, consider the growing popularity of devices such as the Amazon Echo, which many people are now using to control how they watch content on TV. That’s partly because Amazon Echo is so good at finding all the content that’s possible to consume at any moment in time – not just content available on linear TV. And, with Amazon Echo, it’s now possible to do more than just find TV content, it’s also possible to do things like order a pizza during a big ballgame.
Now that we’ve reached a tipping point, where more people watch streaming TV than linear TV, the race is on to find the optimal viewing experience for all those streams coming over the Internet. One of the pioneers in this area is Netflix, which is leveraging innovative deep learning techniques to optimize your viewing experience.
There’s nothing worse than a video stream that pauses or stops during a show, and Netflix is working to correct that problem. Netflix can analyze each frame in a stream and compress it in real time in accordance with the speed of your Internet signal. If you think about this, it’s actually pretty amazing. It means that someone with a very slow, clunky Internet connection will get the same crisp, clear picture as someone with a high-speed Internet connection.
Netflix is also innovating with a new form of narrative TV storytelling that’s much more interactive than today’s TV. In one futuristic scenario, Netflix could give you the power to choose how a show begins and ends – or which characters will appear. In many ways, this is similar to the types of “choose your own adventures” we all enjoyed as kids.
For example, say that you’re watching a Netflix TV show, and it’s obvious that you’re getting bored by it. You could then change the outcome of the show dramatically, to make it more exciting. You might choose an ending where the hero meets an epic demise, or a beginning where the main character is a sexy, young female instead of a boring, middle-aged male. It all adds up to a new form of narrative storytelling. Just as linear TV is going the way of the dinosaur, so are linear TV plots that follow a predictable pattern.
It’s clear that AI and the future of TV are very much connected. The days of the “smart TV” are going to become the days of the “AI-powered TV.” Right now, it might sound like science fiction, but AI is all around us, even today. For example Facebook uses AI algorithms to recognize faces in photos you post online and Netflix uses AI to predict which movies you’ll want to watch. That’s just the beginning – within a very short period of time, the idea of thinking about ways to connect artificial intelligence and your living room TV will become increasingly mainstream.