Can Alibaba AI predict reality TV show winner?

Apr 7, 2016 | Social media

Alibaba has built an artificial intelligence platform that it hopes can correctly predict the outcome of reality TV talent show I’m a Singer. Alibaba’s technology uses performance information such as “voice pitch and energy,” and maps that against factors such as song choice and real-time audience response. The results will be shown online, pitching the […]

Alibaba has built an artificial intelligence platform that it hopes can correctly predict the outcome of reality TV talent show I’m a Singer.


im%20a%20singer%20sm.jpg
Alibaba’s technology uses performance information such as “voice pitch and energy,” and maps that against factors such as song choice and real-time audience response.
The results will be shown online, pitching the technology, named ‘Ai’, against the judges as the show is aired.
The experiment is being held as a “proof-of-concept” for the technology, with Alibaba suggesting that it’ll be used for purposes closer to its core business of online retail in the future.
The Alibaba ‘Ai’ project is being lead by Dr. Min Wanli, the chief scientist for AI at Alibaba Cloud, who was previously part of the IBM TJ Watson Research Center.
Big businesses bet on AI future
This week Google also announced that it was integrating the ability to vote on reality TV shows into its search results.
People can now vote for the X Factor and Got Talent from within search results in Asia.
AI is being increasingly used in entertainment and marketing to assess the return on investment on projects.
Recently, Kang Zhao and Michael Lash, researchers at the University of Iowa, created an artificially intelligent algorithm that helps to determine whether a film will be able to make a profit.
The algorithm uses a variety of factors such as the actors involved, genre and what the film is about and combines them into a massive data set.
The algorithm uses the data to then determine whether a film will be profitable.
Watch this BBC Click video discussing the project:

All topics

Previous editions