Artificial intelligence is being used to figure out how we feel about TV shows, predict plays in sports, and process mountains of data to optimize video business performance.
Figuring out what viewers feel about a TV show
CBS uses thousands of surveys to figure out what is working and what is not in its TV shows. For example, the company conducts over 6,000 surveys when it is testing pilots of TV shows. Processing all the results can be daunting, allowing researchers only to skim the surface of the results.
The company now uses a data analytics solution from New York startup Canvs to help process the survey results. Canvs’s solution employs proprietary AI approaches that allow CBS to pre-process survey feedback on all its programming and special events.
Canvs can automate the interpretation of open-end responses where survey participants give their opinion of a TV show. The ability to parse natural language responses allows the company to spot emotional hot-buttons like love, bored, sad, and anxious. According to Jared Feldman, founder, and CEO of Canvs:
“We have literally billions of comments that reflect the ways people feel.”
Predicting outcomes in sports
One of the major draws of any sport for fans is debating the effectiveness of a team’s performance. AI could help fuel the discussion with a detailed analysis of the game in new ways. Case in point, the Guinness Six Nations Championship of rugby currently taking place in Europe. The tournament is using AWS’s machine learning, deep learning, and analytics services to enhance the on-screen data presented to viewers.
However, Six Nations is going one step further. It is using AWS to perform predictive analysis of real game situations. The tournament games are being tracked in several new ways, including scrum analysis, play patterns, try origins, team trends, ruck analysis, tackle analysis, and field position analysis.
Scrum analysis, for example, will encompass basic statistics like pack weight (the weight of each player in the scrum), player experience, and historical data. It will combine these data with specific performance data from the ruck (when the ball enters the scrum) including speed, cleanouts, steals and more. Analysis of these types of data sets will allow AWS technology to predict in real-time the success of a scrum and other game situations.
If the Six Nations AI experience is successful in predicting the outcomes of specific plays, how long before teams and coaches start using it to improve their decisions during a game?
Spotting trends in mountains of usage data
Digging out information from the enormous amount of subscriber usage data available to video services today is perhaps one of the most productive uses for AI. The technology can be trained to spot behaviors and trends in the data. People can then focus on what to do with the results. Here are a few ways the approach is helping to optimize video business performance:
- Paywizard’s Singula gathers data from each part of the video customer journey and uses it to identify recurrent problems. For example, it can detect an unusually high number of people failing to complete service sign-up and suggest a solution.
- Wicketlabs also gathers data across a service’s many subsystems to calculate a customer happiness index (CHI.) CHI can be used to forecast customers likely to churn and suggest how to stop them from leaving.
- Synamedia uses AI in its Credentials Sharing Insight solution to spot potential abuses of password sharing.