While artificial intelligence (AI) has become well-known in other industries and in other parts of consumers’ lives, it is only just now starting to make a notable impact across the media and entertainment industry. This lag is likely due to the simple fact that video content is complicated.
Once a piece of content has been shot, content creators have a wealth of metadata they can use in developing their programming. Traditionally, this initial collection of data has focused on basic information about the show, such as name, scene, episode, take number, film stop, and codec, knowledge that is critical to current production workflows. Gathering further data about a piece of video typically required manual evaluation of that content, with copious — and time-consuming — note-taking. Few producers were interested in spending money on that process; it simply wasn’t worthwhile.
As a result, media organizations today maintain immense and growing media libraries without really knowing what’s in them. The emergence of AI processing changes that scenario, however. Today’s cognitive engines enable fast, automated examination of files and near-real-time extraction of minute details. These engines can perform processing tasks ranging from transcription, face recognition, and object recognition to sentiment identification, translation, and geolocation. Because additional engines are continually under development, the potential of AI for media likewise continues to improve daily.
Cognitive engines first became available to media organizations as cloud-based processing tools, but their more recent availability in hybrid and on-premise models opens up much greater opportunity for media organizations — particularly those that maintain large media libraries behind their own firewalls. In addition to eliminating the time and cost of moving content to the cloud and back for processing, on-premise examination of local archives eliminates security concerns associated with the cloud. Equally or more importantly, this approach empowers the organization to understand even massive digital media libraries much more fully.
With unprecedented insight into the nature of stored content, media organizations and content creators are equipped to produce stories of far greater depth and texture than ever before. They enjoy greater choice and creative freedom in building a narrative, recounting a history, or documenting an event.
Cognitive engines can identify content that is known to be in the library but is difficult or impossible to track down. In many cases, it was more affordable to go out and re-shoot specific content — such as establishing shots or B-roll footage — than to search for it within a large media library. Now, by using AI, an organization can easily identify essential supporting material within the library and, in turn, utilize it for internal productions or make it available to outside producers as licensable content.
Cognitive engines also solve the significant problem of identifying video content related to a specific person or event. When, for example, David Bowie died in 2016, the world’s leading music television network lacked an effective means of digging into its vast archives and identifying the music videos, interviews, and news stories it had created about the rock star over the decades. Using a cognitive engine for facial recognition, the network could have gone through terabytes and terabytes worth of data in a matter of hours, scanning its archive for Bowie‘s face or voice. With the resulting catalog of available content, building the story of Bowie’s life and legacy would have been greatly simplified.
For news and political reporting, data extracted by a content engine can allow users to search speeches by keywords, faces, and objects; to document various candidates’ and politicians’ positions on key issues; and to track key issues and topics.
Metadata gathered through AI processing not only accelerates the discovery of interesting moments that enrich storytelling but also gives organizations the opportunity to generate new revenue streams from existing content. Quickly gaining momentum, the application of AI to media is proving to be a huge win for media consumers and content creators alike.