NAB 2018: AWS Machine-Learning Tools for Content Creation

At a conference track on machine learning during the NAB Show in Las Vegas, Amazon Web Services M&E worldwide technical leader Usman Shakeel described his company’s toolsets. Shakeel addressed up front the question of whether machine learning can replace human creativity. “Can content ever create itself?” he asked. He emphasized that, in today’s world, machine-learning (ML) tools are being used to create efficient workflows, and curate and extract massive amounts of metadata.

Shakeel reported that ML-aided tools can curate “disaggregated source of metadata.” “Different aspects of the workflow exist across different organizations, systems and physical locations,” he said. “How do you make sure it’s not lost? The key fact is a capable DAM that can manage all metadata across all facets of the media production pipeline.”

amazon_web_services_aws

Machine-learning tools can also aid content security, he said. “We have to look beyond rights management, watermarking and so on,” he said. “They’re important. But it’s important to understand complete workflow security and how we make sure this content ‘lake’ is secure.” Finally, he brought up the “ML-aided content monetization flywheel,” with applications that include “content discoverability and recommendations, personalized dynamic ad-inserts, content viewing experiences, and content production investment decisions.”

“Any metadata that we cover in content curation will help you here as well,” he said.

Using ML tools for content production investment decisions is also becoming more common. “If I’m a studio head, what are the data points I have in making my next decisions?” he asked. “Who is the actor I want to invest in next? Those are important points that our customers are looking for more and more, to bring together a very large swath of users and datasets to make content investment decisions.”

“In summary, there is no Swiss Army knife when it comes to machine learning, especially in content creation,“ said Shakeel. “You have to consider best of the breed toolset for each task and not get stuck with one platform-in-a-box.” He also advised not to move the content around. “Look for a platform that offers a machine learning-aided toolset on top,” he suggested. “From a security perspective, look for a platform that gives next-gen security on top of content. Cloud platform selection will be based on the ecosystem and your partners.”

“Machine learning today is about workflow efficiency,” he concluded. “We have to ask ourselves how to augment the human element with machine learning, and how to build your machine-learning flywheel for your current content business from a monetization perspective. AI is about assisted intelligence.”