An audience of one — served at scale

Media 4.0 is wall-to-wall content, created in real time, on demand, infinitely scalable, cheap, and, most profoundly, serving the needs of an audience of one — at scale. Image from Shutterstock; illustration by Rishad Patel, Splice

This is a scenario Splice was invited to submit to the Open Society AI in Journalism Futures 2024 workshop in April.

Driving forces: A tech-enabled revolution in a distorted media landscape

We are fast entering a paradigm shift in the creation, consumption, and verification of content. Traditional episodic media is giving way to a landscape where AI-generated content is expansive, on-demand, and multi-format. We've passed peak content — AI is creating a content environment of infinite supply.

For Splice, this new change marks the start of what we’re calling Media 4.0 — wall-to-wall content, created in real time, on demand, infinitely scalable, cheap, and, most profoundly, serving the needs of an audience of one — at scale.

This transformation is driven by technological advancements that allow for the instantaneous generation and regeneration of content, tailored to individual needs, languages, formats, preferences, and interests.

The first hint of this came up in Meta’s Q3 2023 earnings call: “Creators will now have all these tools to make content more easily and more fun,” said Mark Zuckerberg. “I think over time, maybe we'll even get to the point where we can just generate content directly for people based on what they might be interested in. That could be really compelling.”

But while technology is the enabler of this change, we cannot ignore the underlying media landscape that facilitated this revolution: a long-tortured institutional media landscape that struggles with a crisis of credibility and relevance with its users. 

That crisis grew from, among other things, an over-supply of content with little heed to actual user demand, a lack of audience segmentation based on user needs, and a failure to structure business models that were able to adapt to new digital formats. 

This was exacerbated by the digital advertising model, which prioritises and incentivises volume (or pageviews) over value (or intent and user relevance), leading to a race to the bottom.

Additionally, there is a disconnect between journalistic products, formats, and outputs and the platforms through which audiences now consume news or information, with social media algorithms skewing the distribution of content away from quality and towards engagement, regardless of the veracity or relevance of the information. 

End state: Serving the decision needs of an audience of one, at scale

The need for information remains the same: we still need to be informed, entertained, make decisions, and be connected. Our individual preferences for how we receive information still vary, based on format, convenience, time of day, phase of life, access, income, and other contextual variables. 

But the atom of what we currently call journalism will change. The current atom is the article (in the form of text, video, audio). The current publishing model is structured around the assumption that democratic values are best perpetuated to a mass audience by knowledge curated by journalists. 

The assumptions around large language models and generative AI are around its now ubiquitous ability to generate content, but the real opportunities for media lie in being able to access and deploy automation functionality and code. 

That means niche (which we define as being specific, not small) is now infinitely automatable.

This creates enormous incentive for media companies to build towards. The ability to discover markets made up of user problems to solve with AI, and therefore real relevance, utility, and value through product-market fit, are endless.

The decision a user needs to make about whether to move closer to their daughter’s school and pay more rent so she can walk ten minutes as opposed to staying further away for a lower rent, resulting in a 3-hour commute is not a story for the Education beat, or a Real Estate podcast. The profound, life-changing decisions that must necessarily be made around this universal scenario shared by millions cannot be — and have never been — served by a headline, two legs of copy, and a grip-and-grin photo. 

The data points needed for this decision need to be contextual, nuanced, and intricate, and depend upon factors like age, number of family members, employment, dynamic population and geopolitical data, and other personalised algorithms that are available only within the context of accessible, infinitely rare human counsel — or LLM-driven AI. 

Imagining the new journalism kitchen 

In the world of the Large Language Model, picture all available data as a warehouse that contains every ingredient in the world, and a dining room filled with people with different culinary needs at different times. Diners are free to specify exactly what dish they want to suit their specific diet, restriction, or format; they are also free to pick from a menu of existing pre-built dishes. Each ingredient and dish is verified by a machine, and guaranteed by a human. The processing of the ingredients into various dishes in different formats happens in the kitchen — wall-to-wall content, created in real time, on demand, infinitely scalable, cheap, and, most profoundly, serving individual diners — at scale. 

Democracy has always been at its best when people are free to access the relevant information they need that enables them to make contextual decisions about their lives. Any model of media, including journalism, that is able to build this into their mission with the use of AI could be well placed to exist in the foreseeable future.

Opportunity through change

We believe that media, especially journalism in its current forms, has one last shot in getting this right. In order to survive, this industry needs to deliver something truly unique, intentional — and centred around the specific needs of its different audiences. 

Using AI to supply more journalism is not the future, nor is it to create demand for that supply. Using AI to find and meet existing demand is.

Being human is the new opportunity

Communities and in-person events that foster human interaction and relationships are likely to increase in value, as they provide platforms for genuine engagement, empathy, and the exchange of human insights that cannot be replicated by machines. 

Platforms like Reddit have built prominence due to their existing ability to facilitate niche-centric discussions and provide human perspectives. Niche subreddits can typically have thousands — and often millions — of users.

Industry Dive built a profitable B2B media business by scaling niche. They built multiple newsletter vertical ‘dives’ — Construction Dive, Education Dive, Marketing Dive, Waste Dive — with dedicated niche audiences. The company was sold to Informa plc, another business that thrives on niche — events.

Value accrues to uniquely human content that resonates on an emotional level, as human curation and insight become a more precious commodity. Human-generated and human-curated content, especially content guaranteed by a human touch (including bias), stands out as a signal of authenticity in a sea of low-value, algorithmically-produced material. Humanness is scarce, and therefore premium.