15 YOUR TAKE Machine Learning START YOUR MACHINE LEARNING JOURNEY
With media companies swimming in content, using machine learning to organise assets
is only a matter of when
research. Here, we can bring best practices from other organisations. The same intelligent metadata can be used for different use cases depending on the functional group. For example, automatically finding text in the lower thirds helps media operations repurpose content for OTT distribution. It can also help a third party subtitling company position overlays. 3. Prioritise and narrow down the list of use cases based on benefits and shortest route to implementation. As part of testing content, GrayMeta helps customers focus on the use case and business benefits rather than just the technology. Taking a realistic and pragmatic approach to what machine learning services work well is key to delivering value. We often need to manage expectations of the capabilities of some machine learning services. For example, speech to text is not 100% accurate today. But the technology is improving all the time – and it’s good enough to deliver measurable benefits to most use cases.
engagements. This not only applies to our machine learning models, but to how we deliver value and increase adoption across different use cases or functional areas. 6. The numbers need to stack up and we help our customers create the business case for wider machine learning adoption. The benefits can be startling. For instance, working with a magazine and radio publisher they identified savings of one day per week for one of their teams dealing with social media publishing. This saving allows them to handle the increase in content expected this year with their existing team. So when you’re walking around the NAB Show and you see signs on a vendor stand mentioning AI, think about how this technology will be adopted within your organisation. There are many successful examples where machine learning is helping content production, operations and distribution today.
START BY TALKING TO PEOPLE IN THE ORGANISATION WHERE METADATA (OR THE LACK OF IT) IS THEIR BIGGEST CHALLENGE 4. Awareness of the fast-moving pace of the field helps us spot opportunities to extend value for our clients. We recently integrated the ability to train the detection and recognition of custom objects into the GrayMeta Platform. Now a branded content agency can highlight where specific products appear in video and images, speeding up the whole production of packages for social media.
5. We use a crawl, walk and run approach, based on constantly testing and learning in our
*Building the AI-Powered Organization, by Tim Fontaine, Brian McCarthy and Tamim Saleh, HBR July-August 2019
feedzinesocial feedmagazine.tv
Powered by FlippingBook