The Undeniable Codependence of Data and Artificial Intelligence

Jamie Mackinlay, CEO

It’s January 2024, and many of us take a moment at this time of year to reflect on the year just past and set goals for the year ahead. For MetaBroadcast, 2023 was a year of embracing the challenges our customers face in understanding the quality of their existing metadata. We launched our industry-leading Audit capability and are actively delivering actionable insights to help our customers optimise metadata costs and improve operational effectiveness.  

As we enter 2024, media organisations and video service providers are clearly seeking options for improving their operating margins. There are typically two ways to do this – drive incremental revenue or reduce operational costs. Given the revenue-related hurdles associated with changing business models, we understand many businesses are focused on cost-cutting. 

With our explicit attention to metadata management, we see how data underlies many aspects of a modern media business. We are not alone, a recent Forrester whitepaper indicated that 77% of media & entertainment decision-makers have made “adopting better data and analytics capabilities” a top priority. 

Data managed by our cloud-based platform, Atlas, is delivered to multiple platforms within our customers’ operations. In our interactions with our customers, we also notice how fragmented metadata management is. Organisational silos have led to data silos. This often results in identical data being paid for more than once and a loss of negotiating power.  It also compels siloed teams to pursue data enrichment when that data may already be available elsewhere within the organisation. The result is data that may be over-provisioned and/or under-utilised, leading to operational drag.

Understanding what data is already being used by any part of the business is the first step towards improved, cost-effective metadata management. The second step is validating its quality. The third step is consistent and persistent management of data to ensure the ongoing integrity of the data underlying internal and external facing systems. 

Our automated processes for validating metadata quality are just one facet of our approach towards the wider integration of artificial intelligence.  At present, we find that data-driven machine learning delivers the immediate value needed by our customers. Viewing machine learning as a subset of artificial intelligence, we see its benefit for metadata quality monitoring. In addition, machine learning algorithms can facilitate metadata tagging and alignment to defined taxonomies. We also see its value in identifying missing data fields and proposing data to fill those fields. In these cases, machine learning assists in maintaining the integrity of the metadata schema and repository.  

We are closely monitoring the adoption of Generative AI and the AI capabilities enabled by our cloud partner, AWS.  Moving forward, we feel Generative AI will be used as a collaborative tool. Our marketing team uses Generative AI for brainstorming but is hesitant to give it full reign in writing newsletters, blogs, or social posts. This seems typical of this moment in time. 

There is valid concern about the ethical and moral aspects of the Generative AI results. As a data-centric business, we fully acknowledge the challenge of “garbage in, garbage out” when it comes to data or, as was mentioned recently, AI-generated hallucinations. Are the results based on facts or bot-generated opinions? For now, we will pursue a pragmatic approach such as human-assisted generative AI.

We feel there are several use cases where AI will facilitate metadata management.  These range from enhancing data quality to data creation (e.g., suggestions for synopses describing content assets or auto-generation of tags or keywords) to localisation (e.g., auto-translation and event markers). With every use case, there is more data, with estimates that AI has increased the volume of metadata by more than 300%.

Now, more than ever, metadata management is vital to ensuring the accuracy of the underlying metadata. Our customers cannot succeed without high-integrity, high-quality metadata. With this in mind, we will adopt artificial intelligence where and when we are confident that neither integrity nor quality of data are at risk. 

We look forward to aligning with our customers and partners as 2024 progresses. The codependence of data and artificial intelligence holds great promise but must be approached with a pragmatic balance of innovation and vigilance.