AI in practice - Olympics Daily Recap
Earlier this week I attended a Datastax event in London focused on building AI applications.
One of the speakers was Patrick Miceli, CTO, NBCUniversal who talked through their use of AI as part of their Olympics coverage.
The Olympics presents an interesting broadcasting challenge in that over the course of two weeks there are thousands of events, with hundreds of individual events taking place each day.
The problem statement is how do you help each viewer get a valuable recap each day that reflects their personal interests - as not everyone is interested in the same sports.
The NBCUniversal team set out to create a personalised daily recap powered by AI.
Hosted on their Peacock streaming platform, each day the AI would craft a personalised and unique 10 minute highlight reel.
It was based on the viewer opting into the types of highlights they would would like to see, including three favourite sports and preferred topics, such as behind-the-scenes and backstories, top competition, viral and trending moments, and a spotlight on international teams.
The platform incorporated the AI-synthesised voice of US commentator Al Michaels to give your own recap a professional and real feel.
Perceive, infer and synthesise
What NBCUniversal have tapped into is using AI to build a new connection between two things they already knew:
They knew all of the sports events that had happened the previous day (because they had the rights and had streamed them all)
They knew the viewers personal interests, from what they had selected and their viewing habits.
AI was able to perceive these two data sources, to synthesise them, and infer the most appropriate 10 minute highlights real.
How does this apply in your business?
Data is the fuel for AI. But too often we think about data as something that we already have access to - instead of thinking about new data or knowledge, that if we created or uncovered it, could be of value to our customers when combined with other data we have access to.
Do you capture your customer’s product usage, but not do anything with it today?
Do you capture details of all your customer's’ different users and their teams, but not do anything with it today?
Do you have aggregated knowledge of how customers in different industries, company sizes or countries use your product?
Do you know how the questions that a customer asks during a sales process might impact their adoption and expansion in future?
Do you know how weather, the economy, or demographics affect your customers’ use of your product?
Bringing two separate pieces of information together can be the catalyst for new and improved value propositions for your customers.