Harness GenAI for Disruption in Content Production

GenAI is a double-edged sword, representing both a risk and a massive opportunity. As the cost production for content trends to zero, the importance of journalistic storytelling, information sourcing and brand voice increase in value. So… How are the best media groups using GenAI? (This article is republished with kind permission from the original in the Media Producers’ Edge by www.outerop.com)

First, a quick data detour. Over the past two if not three decades, the internet and social media have disrupted journalism (duh), causing a 52% reduction in revenues for newspaper publishers. That’s a lot!

The internet and social media disrupted individual publishers by aggregating, ranking and personalizing news feeds for end-users. So it’s no surprise that with social media adoption in the US reaching 70% of the population, an estimated 50% of US adults say that they now get news from social media channels.

The hard part of media production is in “production”, notably, creating the content. Once this has been done, then distributing it via additional channels seems like a lot less work. Yet that does not reflect our review of media publisher’s social data.

There’s a huge range of social media presence across media firms:

CNN are doing YouTube right. This short, for example, has 204,000 views, content which has been repurposed, which means that it’s doing double dipping correctly. See LINK

But the future may be slightly brighter with GenAI as a possible production equalizer

Right now the average time to write an article is 2+ hours, Summaries and metadata for search traffic may add on 20 minutes. And that’s not considering video broadcasting and production, which can run into tens of thousands of dollars per hour.

The competitors are clear: it’s independent journalists weaponized by Substack, beehiiv, YouTube and TikTok. The last remaining differentiation of media production is journalistic voice and storytelling. The careful integration of GenAI into journalistic workflows can help for three key reasons. It can help:

1. Create more high-quality content via efficiency gains.
2. Better target advertising and increase revenues
3. Increasing engagement more effectively

Content Creation Efficiencies

At the start of a content workflow is story sourcing. This is hard to re-create. Media organizations have spent decades nurturing relationships to have differentiated access to sources. Most already use GenAI with social listening to find and surface relevant events and details from social media sources.

Next comes content drafting. It has been the obvious usecase for GenAI but there’s a balancing act between automating content creation and people-led journalism. And to be honest, this is not new…See The Washington Post‘s AI Heliograf, a “robot journalist” for short articles + local events published in 2016.

And Forbes x AI = Bertie an AI powered CMS in 2018

But this wave of AI is different. What happens when GenAI can do (or assist) the things that traditional human editorial staff excel at. Check out this quote from Wired about Reuters 2018 AI initiative Lynx Insight:

“Reuters says the aim is to divvy up editorial work into what machines do best (such as chew through data and spot patterns), and what human editorial staff excel at (such as asking questions, judging importance, understanding context).”

We expect co-pilots in media to go further to assist digital producers and journalists than ever before. This is an opportunity to scale sourcing with AI that can ask questions and judge importance by understanding larger amounts of data and context. This is the challenge that all media companies are facing.

So how can we add GenAI into our processes whilst maintaining our editorial quality? And how can one achieve a balance between human writing, which has brand value and story-telling excellence, and automation?

We believe that the best examples of integration treats AI content like an initial draft which needs a human review before pushing out. Éduord Guihaire at Agence France Presse (AFP) says it best: “It’s important that journalists become familiar with these technologies, use them, test them out, and consider them like a suggestion box,” he explains. “But it will always be humans that supervise, check, and validate.”

Thus, control should be part of the editorial process to ensure that the content created is correct and in the right brand voice. (AI hallucinations (erroneous summaries) are a problem). Content drafted by AI is notoriously hard to align with the actual “brand voice” of the journalist or media companies. However, techniques like prompt optimization and fine-tuning can help solve this to elicit a specific tone-of-voice.

Better ad personalization

Specific and personalized ads sell more effectively. Period. And guess where most of media groups income is from? Yep, advertising. (Editorial Note: At the same time, for certain print and online magazines, such as The Atlantic and The Spectator, it is paid subscriptions, with advertising representing only a small portion. It is a completely different story with newspapers such as The Daily Mirror where advertising is critical). The best of the best media groups create data pipelines, combining content tagging and user behaviour to enable dynamic advertising i.e. matching advertising to more effectively convert users.

Ever wondered why adverts seem to know you better than you know yourself? Dynamic advertising that’s how. Media organizations which require sign-in have the strongest metrics on customer interests, alternate privacy-compliant IP and device tracking can assist with user behavioural tracking too.

Here GenAI is being used to:

1. Correctly understand or tag content so that advertising can be matched with the visitor’s interests more efficiently
2. Dynamically create different versions of advertising (copy, images and soon videos) to share at the perfect time and unique content

The more recent unlock is for videos and podcasts. GenAI can be used to tag content at scale, e.g. analysing video transcripts. Language models extract key topics and emotions, thus enabling better targeting of end customers, all of which results in higher return on ad spend (ROAS) which is more attractive to partners.

Increasing engagement more effectively

Engagement, engagement, engagement. Other than bringing amazing stories, facts and news to the world, this is what media businesses live on. We feel that brands have not been getting the most out of social media. This is partially because re-creating content for short-form channels (e.g. YouTube, TikTok, Instagram, Facebook) is like doing the work twice!

GenAI has emerged as the “production disruptor”. It speeds up the ability to produce short-form media and to re-purpose that golden, golden content across channels. This doubles or even triples the engagement from the original source. With algorithmically driven social feeds, repurposed content CAN be viral content. Nice one BBC! (See LINK)

Massively increasing engagement to a channel while expanding ad revenues on social media channels can be juicy for media brands, with the GenAI cost reduction it can provide a strong ROI and teams can do more with less. From content repurposing to AI voiceovers, this AI wave is the gift that keeps on giving. Broadcast media companies can leverage GenAI tools like Opus Clip to re-purpose content and ElevenLabs to easily voiceover content and share across multiple languages plus automated captions, generating increased engagement.

As social media continues to disrupt traditional journalism, traditional media has a potential knight in shining armour in GenAI

By embracing technology, traditional media can create more high-quality journalistic content via efficiency gains, better targeted advertising to increase revenues, and enhance engagement effectively. GenAI enables faster production, improved content personalization, and targeted advertising, driving both engagement and financial performance. Ultimately, it’s about leveraging it to regain a competitive edge.

Pavel Patel and Stephen Byrne are the founders of Outerop, a GenAI development platform. Stephen was a VP at Fox, while Pavel used to help run a Series A start-up.