Machine Learning And AI In The Media Industry

Michael Brenner

We live in an age in which content is king. Take a look at the media landscape in 2017:

  • Newspapers and magazines have given way to online platforms and social media news delivery
  • Linear television channels have been blown wide open, with viewers enjoying their favorite shows and movies, wherever and whenever they decide
  • The days of the single-television household have been replaced by the age of multiple devices, of simultaneous viewing across multiple platforms

But what does this mean for both content consumers and content producers? It means more choice, it means more flexibility, it means that, if the content is no good, the consumer will simply go elsewhere. It is this level of competition and anxiety over consumer habits that has led to the Golden Age of television… the age of television as high art.

This is not the only result, however. The non-linearity of modern media production means that content producers are not competing for a limited number of slots. There are numerous channels and on-demand services in which producers can distribute their visual content, and even more for music producers.

Which begs the question: In this increasingly diversified media landscape, how do consumers find the content they really want?

Artificial intelligence and the paradox of choice

Artificial intelligence could provide the answer. By providing the right data fields, we can help content delivery platforms develop an understanding of what sort of content we respond to best. These platforms can then recommend content based on this knowledge of what we like.

This is a technological response to what psychologist Barry Schwartz described as the Paradox of Choice. The theory states that there is a tipping point for choice; a point at which choice ceases to provide an advantage and instead becomes a hindrance.

For example, we may be energized by the thought of thousands upon thousands of movies to choose from, but anyone who has spent an unseemly amount of time scrolling through Netflix will understand that this is not necessarily the case.

Through machine learning and exploiting consumer data, we can bypass this, connecting with great movies, shows, and other content that we may otherwise have missed.

Can machine learning go deeper?

The situation described above sounds pleasant enough, but is it really getting the best out of machine learning and AI? It is great that we can receive the content we want, when we want it, but this is simply an extension of technology that already exists.

Every time we shop on Amazon, every time we are presented with an advertisement on Facebook or Instagram, every time we use a search engine like Google, we are witnessing this process in action. Algorithms utilize data you have provided from previous activity and from data capture forms to deliver to you what the system “thinks” you want to see.

Building this into a content delivery system is a little more impressive, but, still, can’t we do more? The answer is – as it so often is – yes, of course we can.

Effective social media analysis

If machine learning is to provide an effective antidote to an increasingly saturated content market, it needs to truly understand. This means gaining real-life insights that go far beyond such simple summaries as “You watched Jurassic Park, maybe you’d like to watch The Lost World?”

For this, we can look to social media and the sea of data that exists online. This is the 21st century’s water cooler; this is where conversations are had, tastes are made and broken, this is the furnace in which the classics of tomorrow are formed.

What makes social media so beautiful for human users is also what makes it so awkward for machines: the conversation is natural. If you’ve ever encountered a bot account on Twitter, you will know what I mean; faking human interaction is very difficult.

But we are not talking about faking, here. We are talking about understanding.

In 2017, Crimson Hexagon published an article on this topic, but with regard to corporate branding rather than content distribution. Writer Garrett Huddy discussed how many organizations use familiar words and terms as brand names – Huddy used the example of Tide detergent, but there are countless others, including Mars, Apple, and Coach.

  • “I love my athletic coach’s shoes today, I want to buy some.”
  • “I love the athletic shoes in Coach today, I want to buy some.”

Two very similar statements, with two very different meanings for Coach’s marketing department. Modern machine learning is about far more than just recognizing words; it involves complex analysis of word patterns and grammatical structures, as well as context. Only once all this is taken into account can the system decide if the insight is relevant or not.

This works for media products, too. By understanding the conversation and the response to media products, machine learning systems can gain deeper insights into what sort of media we want to see.

How does AI support the media industry?

So, how is this helping to support the media industry? How is this helping generate revenue so that production companies can keep on giving us the music, movies, and TV shows we love?

Moves away from physical media products – such as CDs and DVDs – to streaming services have created a headache for filmmakers, musicians, and production companies. For a little while, it seemed like traditional monetization of media products was doomed, but through advances in machine learning, the revenue streams that support artists are being secured.

Software such as Pippa enables podcasters and live broadcasters to insert ads into their offerings, using AI to determine the ideal ad content and deployment. Jaak, on the other hand, uses blockchain technology usually associated with cryptocurrency networks. This enables content producers to know exactly where and when their work was streamed and by whom. By integrating with a digital currency network, this innovation supports instant payments for producers, with no latency period.

Artificial intelligence technology can also be used to analyze the appetite for different kinds of art and media and apply this to new, up and coming artists. By identifying musical talent early, production companies can secure the future of the next generation of art, media, and culture.

Learn about SAP Brand Impact, a new AI analytic solution that applies machine learning and artificial intelligence to high-level content analysis.


About Michael Brenner

Michael Brenner is a globally-recognized keynote speaker, author of  The Content Formula and the CEO of Marketing Insider GroupHe has worked in leadership positions in sales and marketing for global brands like SAP and Nielsen, as well as for thriving startups. Today, Michael shares his passion on leadership and marketing strategies that deliver customer value and business impact. He is recognized by the Huffington Post as a Top Business Keynote Speaker and   a top  CMO influencer by Forbes.