How will AI technology impact editorial and content creation?

Follow Up Article from Tony Treacy at Evolved Media

There has been much speculation about the impact of AI on content creation, much of it pointing to AI tools replacing journalists and writers, and job losses from automating manual tasks. But how much of this is true? How do AI tools really work, and how can journalists, writers and content creators leverage these new tools to produce digital and analogue content more effectively?

Russell Pierpoint, Managing Director at Evolved Media, and Ben Martin, Managing Director at True 212, ran a breakfast briefing to address this question on Thursday December 5th at the PPA offices in London, and below is a summary of what they talked about.

Russell introduced the briefing and began by explaining what AI means. He said: ‘AI generally means computers doing things that traditionally humans would do.’ He went on to explain how there are two parts to AI. AI is the result that people see but it is underpinned by machine learning. This is the analysis and processing of data to identify patterns and objects, the results of which are then fed back.

Although, some people may think AI is something for the future, there are already numerous examples where AI is being used. Social media platforms and websites suggest content to users all the time, which is driven by AI. Voice recognition AI in services like Google and Alexa perform tasks as instructed. Face recognition is well known in Facebook.

However, publishing (and journalism) is still in the early stages of adopting AI. Ben referenced the Polis Report: ‘New powers, new responsibilities. A global survey of journalism and artificial intelligence’, launched in November, which estimated the real impact of AI on journalism would happen during the next two years. But at the same time, analysis found that 63% of news organisations did not have an AI strategy in place.

In Ben’s opinion there are five main reasons why they don’t have a strategy: “I think you can point to a lack of financial resource, skills, knowledge, strategic thinking and degree of scepticism about the value of AI” he said.

For Russell there is also a clear and obvious gap. Not all publishers have implemented technology to fully manage legacy content, and the editorial and content creation processes. As Russell explained: “You need to have the right technology managing your content in order to plug in AI tools. Once you are connected you can then expose your content and data and start to see the benefits. Without the investment in the core technology, how can you make a plan to use AI tools?”

It’s not that there is a shortage of tools. A whole range of AI tools have been developed by Google, Amazon and Microsoft, and there are also a number of specialist AI services that have been developed by niche software companies, including Clarifai, Site and Musimap (to name three examples). Amazon probably offer the most comprehensive suite of tools, as Russell explained: “Amazon are very good at turning AI technology into products and making them available to businesses. They currently have 170 AI services available, and at their recent conference in the US they announced a further 30 new services that will be coming to market soon.”

In terms of publishing and journalism both Russell and Ben see AI impacting editorial processes, content creation and production, and distribution, but for this meeting they focused on how AI aids content management (of images and video), and journalism / copywriting.

Tagging images (and video etc.) is probably the most commonly used AI tool in publishing today. Russell demonstrated how this works using Wooding’s Elvis DAM system connected to Amazon Recognition and Google Vision. As images were uploaded to the asset management system, the AI app provided a list of tags for each image that could be added to the database if required. Russell described how in recent years the quality and richness of this information has significantly improved as the technology has developed: “As it has learned how to identify objects in images more accurately,” as he explained.

The images uploaded in the demo were wide ranging, and it was clear to see how the different technology platforms returned different but meaningful tags. It was also clear that tagging has moved from basic descriptions, for example recognising a car or its colour is red, to recognising the car was, in fact, a Mustang Fastback.

The benefits of tagging images, whether at the point of ingestion or as part of a data enhancement project later on, are clear from Russell’s perspective: ‘Having a database of images, graphics or video is clearly a valuable asset to any publisher. However, the challenge for editorial, design and production is finding the image or video they want. The search capability of any database is determined by metadata, and because each department might want to search for very different reasons, the metadata needs to be as rich and broad as possible. AI tools generate really useful metadata automatically that could not be entered manually cost-effectively.”.

So, tagging images is an extremely worthwhile exercise that in Russell’s opinion will aid search and save time, and as Ben pointed out, that might be enough ROI to justify the investment in the first place.

Ben’s company are not only users of AI technology, but are also in the process of developing a new editorial tool through new a division called ‘Loyal AI.’ He demonstrated one of the Twitter social media tools live, working in Google docs and also in Woodwings’ Content Station platform. The ambition, overall, is to provide journalists and writers with faster search, fresh perspectives and greater engagement, both for them as writers, and ultimately for the end reader. The Twitter tool provides insights for journalists in real time, enabling them to find out more about subjects, people, places and events as they write.

The benefits, as Ben explained are: “We want what journalists write to be more engaging for the reader, who will therefore spend more time reading it (increased dwell time), and we want to improve the content creation process, by making it more efficient. Our target is to speed the editorial process up by 20-30%.”

Ben is looking for partners in January 2020, to help test the new tools so if you are interested please contact him at ben@true212.com

If you are considering how best to augment your digital asset management content, or want to talk about developing a more effective technology platform capable of connecting to AI tools, please contact Russell at russell@evolvedmedia.co.uk