AI For Business: How To Train Your Algorithm

Saïd Business School
9 min readFeb 1, 2021

This year we launch our new Artificial Intelligence for Business diploma at the Saïd Business School.

With academic and industry experts, we’ll be looking at the fascinating implications and opportunities in AI and its associated technologies — machine learning, neural networks and so forth — for all kinds of industries in the 2020s. We’ll consider government, intelligence and information war, financial analysis and trading, technology in general, medtech in particular, law and ethics, agriculture and many more areas. As one clue to our thinking, and the radical dimensions of the change which is sweeping through every sector, let’s have a look at just one sector — and how much it has metamorphosed: the media.

Of course, like in a science fiction film, the lockdown of 2020–21 already transported almost everything in the media into a hitherto little-imagined new dimension. Cinemas closed globally, Manchester United played to an empty stadium, a new, $1bn James Bond film sat on the shelf, for a year. But more important still, the media business found itself in a universe with new laws. Artificial intelligence shorted out the established business models like the opening sequence of Stranger Things. The determinants of success, from media buying in advertising to set design and build in TV drama, became defined by a new paradigm which we might codify in a simple phrase: how to train your algorithm.

Media production, distribution and monetisation has been impacted by radical new technologies — like artificial intelligence and ‘black box’ neural networks, or unreal engines in content creation — neutralising board room playbooks, and rendering a library of university textbooks obsolete. From 2020 onwards, the winners in content distribution (think Netflix personalisation), production (virtual sets gave The Mandalorian its premium impact), music (Spotify’s peerless music recommendations), gaming (now $200bn+ annually), and even events and newspaper publishing (the FT and New York Times are years ahead on analytics) have been defined by their ability to deploy tools like artificial intelligence to match individual audience members to content and adverts.

So let’s look at some indicative examples of where we are now in media, and where we are going with AI — as an example of how deep we can go on our course into any sector.

We’ll do it across a set of categories: retail and content, advertising and TV, streaming and TV production, and live and interactive. To our own intuitive daily experience, every one of these areas has been re-mapped by AI.

Retail and content

Every retailer now needs to also be a content company, and every content company a retailer; you can see it in your Instagram feed, or Rihanna’s $1bn+ Fenty brand. Content makes you swipe to buy, and ecommerce companies use it to drive sales more cost effectively than Google ads, where customer acquisition cost is high. Retailers are using their financial power to aggregate tens or even hundreds of millions of followers, disintermediating the content producers of old — like commercial TV — and adrenalising direct sales.

How? They are using AI in targeting and distribution of that content, A/B testing at scale, and using dynamic content optimisation to test many variants of any given piece of content on their websites to make sure that the perfect version is available to each individual viewer — the right language, length, colours, price and many more parameters.

Booming ecommerce companies such as Germany’s Zalando or Metro, or UK unicorns such as Birmingham’s Gymshark (valued above £1.2bn in a recent financing), and Manchester’s THG (listed at £7.7bn) and BooHoo (£4.51bn) all added content studios in recent years, from stills to social video, commercials and films. THG is even launching Icon, the size of a major Hollywood sound stage, behind an Amazon distribution warehouse, and next to a major freight airport in Manchester — to create smart content and use smart distribution to reach its worldwide audience with the right video at the right time for each customer.

Amazon are integrating direct shopping into its own shows, on Prime Video. Social platforms Snapchat, Facebook and Instagram are massively upscaling their ecommerce offers. The Chinese model of social content and ecommerce combined is showing the way for Europe, with a group buying function like Pinduoduo. Wherever you look in retail, content and AI are redefining the shopping experience.

Algorithmic advertising to the rescue of commercial TV

There is a similar revolution being played out in TV. Faced with existential challenge from the streamers like Netflix and Disney+ — with their AI-powered recommendation engines far outperforming on personalisation of both content and advertising — broadcasters are joining them in 2021. Commercially funded (though public-owned) Channel 4 has now repositioned itself in 2020 as the UK’s largest free streaming service. It is advertising-funded, not subscription-funded like the big players, but aims for 40% digital revenue by 2024. The BBC, founded 90 years ago, in late 2020 dissolved its hitherto-important channel controller roles, giving commissioning power to iPlayer-first genre heads instead — like Netflix. Yet such is the pace of change that less than 15 years ago, Netflix was a videos-by-mail distribution company, which Blockbuster Video declined to buy.

For TV networks facing existential challenge to their model and programming budget cuts, one AI-powered tool now exists to finesse the jump from broadcast to streaming, and from brand to performance. This will be what in the US has been called a-commerce, or algorithm-driven ecommerce advertising. It has the targeting of Google ads, the contextualised content of social ads, but the production values of proper TV. Their success in doing so will drive the long term commercial futures of the major commercial broadcasters worldwide.

But commercially, the challenge has only just begun, as this Broadcast Video on Demand (BVOD) model seeks to counter both the subscription, ads-free model of Netflix and Disney+, and the narrative that broadcast advertising and its brand-building capabilities are a luxury that precision-driven performance marketing can systematically beat.

Streaming and dazzling new production methods

AI is everywhere in the making of content too. It’s the excellent lighting algorithms in the iphone 12 which mean it is increasingly being used by TV producers to shoot 4k video for broadcast. Or creatively, what made Disney+ series The Mandalorian such a leap — it became the US-top-rated streamed title of Christmas 2020 — was Stagecraft, a new technology. Like the Unreal Engine which drives modern computer games, it renders three-dimensional photo-real backdrops to move perspective with the camera position. The reward ? An astounding ten-series Star Wars order.

Streaming is peak customer centricity: a perfect TV restaurant, in which the waiter has guessed each diner’s order without even showing them the menu. Binge engagement feeds lower churn and marketing costs, higher gross profits, more content investment and happy shareholders. All of that is powered by AI.

Human plus AI in content creation

If we had to put a sub-title on our Oxford AI for Business course, it might have been this: ‘Human plus AI beats AI alone.’

Data and AI-driven recommendation power Spotify — but that’s not everything you need in media. The instinct of a great producer will have more value than ever, especially when combined with statistical insights. As Bela Bajaria, head of Netflix global TV observed: “‘There was never any data that would have proven that a series about a female chess prodigy would have captured hearts and minds in the way Queen’s Gambit eventually did.”

Is there a contradiction in that? Not at all. Artificial intelligence company Deepmind, part of Google, found that Human plus its AI computer AlphaZero beat AI alone at Chess. It is the same in media. So competition is more avid than ever for the services of the elite producers who can deliver hit drama and reality series — and not just in the English language.

Live & Interactive

We know the epic costs of lockdown, human and financial. But the new lived experience has created winners too. Epic Games’ Fortnite broke all records — like 12.3m concurrent players watching a Travis Scott performance, a true waypoint in the reinvention of music marketing.

Everywhere it can, interaction, some of it AI-powered, is growing. Live streamer Twitch trebled its active channels in 2020, taking it to 5.1bn hours watched in Q2, and making the billion dollars Amazon paid for it in 2014 a media asset bargain. Activision’s blitzkrieg shooter product Call of Duty was downloaded almost a million times per day in 2020 — just on mobile. Its interactive livestream became a thirtysomethings social safety valve, like the weekly ‘Zoom drinks’. Games distributor Steam, even without the algorithmic recommendation layer powering 2020s ecommerce, added 2.6m first time buyers each month of 2020, to go to 120.4m active monthly users. And now it’s about to launch in China.

Hardware grew too. Sony and Xbox face vertical demand curves for new console generations. UK-listed Games Workshop can’t produce its role-play toys fast enough, maxing out its factories and more than trebling its market capitalisation across a year, to £3.5bn. Even also-ran technology Virtual Reality finally took off, with 71% more game sales than 2019, a breakthrough game (Half Life: Alyx) and new kit that’s actually fun for interactive gaming, the Oculus Quest 2. AI offers countless advantages to VR, from sophisticated avatars engaging in realistic in-game conversation, to clever tracking of user behaviours through the six-dimension inputs of a VR headset and handset combination offering character, response and even medical user profiling.

(Note: the ethics of that are yet to be explored but will be a huge issue as the 2020s goes on. If Oculus knows your instant physical responses at a nuanced level to individual types of content, that would be powerful psychographic data for advertisers, but as Cambridge Analytica showed us, its use may not always be ethically sound.)

Events growing bigger than ever — and powered by virtual studios and 5G

If a tidal wave of pent-up demand is a critical sociological facet of the post-covid era –shows and festivals will benefit, Technology has moved forward in the events business. This is not the poor facsimile of events on video conference– January 2021’s Consumer Electronics Show achieved a tiny fraction of it’s usual PR buzz, proving that the physical locale of Las Vegas is still critical to its event DNA. This is actual live events equipment that will affect everything from product launches to trade events from late 2021 onwards. And it’s AI powered.

The Unreal Engine, the 3D creation tool behind the new generation of games, creates immersive worlds in the 2D space (ie without you needing VR glasses) and that has huge event applications. Expect a wow factor on big screens at even the most trade-focussed of B2B shows. Combine that with the real-time, no-latency, two-way bandwidth of 5G networks (set to finally hit meaningful speeds in Europe and the US), and the virtual studio screen technologies seen in the new generation of TV studios will drive new must-see-in-person experience that actually have a participative element at events. Via your phone, you’ll be on the big screen, as well as watching it.

And again, as 5G arrives, expect much more crowd interaction on the big screens, including rendered 3D concert imagery streamed from thousands of phones using localised 5G transmitters. By mid-decade, you should be able to watch a concert from any angle of your choice, by virtue of thousands of synthesised 5G livestreams from user phones. AI will be a critical technology in the delivery.

So in just one field — media — AI is having dramatic effects on every stage of the production distribution, and consumption process. Across our Oxford AI for Business course we will go deeper, look at the technology, and go broader, looking at every sector. We look forward to you joining us for this incredible journey.

WORDS: Dr Alex Connock, Co-Director, Oxford Artificial Intelligence for Business programme 2021

PHOTOGRAPHY: Shutterstock

To find out more about the Oxford Executive Diploma in Artificial Intelligence for Business:

Download the programme brochure

Visit the programme homepage

Arrange a call to discuss your eligibility



Saïd Business School

At Oxford University’s Saïd Business School, we create business leaders who lead with purpose.