AI in TV: Service providers seek proof that AI works across the TV UX

AI may touch about every aspect of the television user experience but content providers are increasingly demanding evidence that it really does deliver on all those promises from vendors of the technology.

In this article (CSI Magazine / April 2020 edition — page 18), Philip Hunter assesses what new applications the use of AI opens around the television user experience and how it can enrich video.

Jacques-Edouard Guillemot contends that AI applied to poor data merely amplified the flaws and would be totally counterproductive. “Not a lot of people are doing it yet. You have to have to have data of good quality, people who understand data and processes to exploit it. That is 90% of the job, and AI or Machine Learning is just the icing on the cake.”

Guillemot reckons there is still a lot of money to be earned by just applying the basics of data preparation and analytics without that icing. That said, he acknowledges the potential of AI in various departments. One of those lies in the area he emphasised; preparation and normalisation of the data itself.

“When you have harmonised clean data between CRM, consumption and content metadata, then you have more detail about location, revenue and other key measures,” says Guillemot. The next step might then be to enrich the metadata to provide a stronger foundation for recommending content by linking deeper knowledge of the asset with the affinity of subscribers. “But there is no evidence of that happening yet as most operators are still trying to establish basic metadata.”

Content metadata, however, is only applicable to on-demand video where there is time to generate it before distribution. Advance recommendation on the basis of prior knowledge of the content is irrelevant in any case for live sports where viewing decisions are based on user expectations rather than information. Partly for this reason, but also because metadata alone is insufficient to generate effective recommendations even for VoD content, operators need additional tools in their armoury and this is where AI/ ML can come in.

Continue reading this article (page 18): https://www.csimagazine.com/eblast/Digital_Editions/April2020/CSI-April2020-DigitalEdition.pdf