Towards sustainable publisher models for the AI-driven information ecology

The creation of an AI-driven information ecology

For most of the modern era, news publishers have had a fairly stable and easy to define role in society: Journalists gathered information and produced stories, editors bundled them together, publishers distributed those stories in newspapers, broadcasts, websites and apps.

Since the turn of the millennium, digitisation has chipped away at that model – and in the coming decade artificial intelligence (AI) is poised to transform the model in fundamental ways.

The change is often perceived as a threat to news publishers. Declining direct traffic to news publishers, misinformation and fake news, job losses from automation, and the erosion of trust of news. All these developments matter and must be taken seriously, but they describe only the destruction of the existing model. The even more important question is: what AI is creating?

The transition to an AI-driven information ecology is not simply another technological disruption of news publishing. It is a fundamental reorganisation of how information is produced, maintained, discovered, interpreted, and used to create value. In the next five to ten years, AI will not only change how journalism is produced but how information itself flows through society. This will create a new information ecology of agents and flows that is likely to gradually displace the information ecology of articles, websites and apps that we know today.

When this happens, the central strategic challenge for news publishers is not how to use AI to produce and distribute news more efficiently. It is how to redefine their editorial and economic role when static articles, videos or podcasts cease to be the main unit of information exchange, and AI systems increasingly mediate the relationship between information and its users.

In this essay, I outline a positive vision for the strategic position of news publishers in an AI-driven information ecology. I argue that the central opportunity for news publishers is to transform themselves from producers of media products into companies that originate, maintain and commercialise trustworthy information across multiple markets. This vision does not transform the core mission of news publishers. Rather, it resituates the editorial mission in an AI-driven information ecology and points at ways to realise the mission in an economically sustainable way.

The vision is not a prediction. It is an ambition or strategic goal that publishers can strive for. Further, it is based on my best guess of how the AI-driven information ecology will develop. This guess is inspired by the thinking of Shuwei Fang, David Caswell, Ezra Eeman, Florent Daudens and others, but represents my own synthesis of how publishers might position themselves in an AI-driven future.

The AI-driven information ecology emerges

Several times in modern history, technology has changed how information is produced, distributed and used. In the analogue information ecology, information was packed into finished media products such as newspapers and broadcasts and distributed to mass audiences. In the digital information ecology that emerged in the 2000s, information became digital objects such as articles and videos that were accessible through networks, and increasingly controlled via social platforms and search engines.

AI is now making a more fundamental transformation possible. AI systems are increasingly producing, retrieving, and synthesising information for specific purposes. Individuals are progressively experiencing information through AI systems that select, combine and interpret information on their behalf. Companies are now using AI agents to discover and analyse information that supports their operations and decisions. Communities are developing new information environments to knit them together and set them apart.

The deeper transformation is that the basic unit of information may shift from finished media products and static digital objects such as articles and videos, towards more fundamental units of knowledge that can be recombined to serve different purposes on demand. This makes it possible to separate the production of information from its uses. The same information can flow into numerous services and applications, creating value in different ways that are adapted to the intentions and contexts of individual active or passive users. In the coming years, this transformation is likely to deepen and widen.

As of now, this emerging AI-driven information ecology exists in parallel to (and to some extent, builds on) the traditional digital ecology of websites and apps that rose in the 2000s and matured in the 2010s, gradually displacing the analogue ecology. In the coming years, the emerging information ecology will grow relative to the traditional ecology and eventually become the primary mode for information to flow. Table 1 describes core dimensions of the three successive information ecologies.

How AI reorganises the information value chain

Figure 1. Two-way information value chain in the AI information ecology

In the first step of this transformed value chain, new information is added to the information flows of the information ecology as real-world events unfold. This is done by machines (e.g. via sensors) as well as by humans (empowered by AI), as some information production continues to require physical presence, as well as access and emotional connections to sources and situations.

In the second step, the produced original information is organised in a knowledge infrastructure containing the accumulated information that has been produced over time.

In the third step, information from the knowledge infrastructure is orchestrated to address specific information needs. The orchestration is carried out by AI systems that operate within guidelines and guardrails set out by the companies or institutions that control them.

In the fourth step, the orchestrated information is presented to users. This happens in AI services that provide answers, understanding, advice, decisions, completed tasks or actions – whatever is actively or passively requested by users.

This proposed new information value chain describes economic activities that can function separately as well as upstream and downstream information flows. Downstream, information is added, organised and transformed to fill specific needs. Upstream, signals about the information needs of users are channelled, informing the orchestration of information synthesis and providing data (new knowledge) that is in demand.

The economics of the new information value chain

The example above illustrates how the same piece of information can create value in different contexts as it flows through the new information value chain. But it does not answer the fundamental economic question: Who creates value and who captures it?

Three developments are likely to shape the economic dynamics of the new information value chain:

AI expands information markets: AI is poised to expand the market for information by reducing the cost of producing and transforming information and increasing the number of purposes it can serve. This creates new information markets along the value chain that publishers and other information producing actors can participate in and (potentially) derive value from.

AI creates new scarcities: The growing abundance of information changes where scarcity – and thus, economic value – is situated. While the types of information that can be produced by everyone with AI are likely to become abundant and commoditised, two types of information are likely to remain scarce. First, original, trustworthy knowledge that AI systems cannot independently produce is likely to become more valuable as AI increases its uses. Second, access to the intentions and context of users is likely to become an increasingly valuable resource, as it determines what needs to be met.

AI shifts bargaining power: The expansion of markets and creation of new scarcities shifts bargaining power towards organisations that control the new scarcities: Original knowledge and information about user intentions and contexts. Accordingly, organisations that control such resources are likely to be able to capture a large portion of the value created in the AI-driven information ecology.

While there are strong grounds for expecting that AI will expand the information economy and increase the value of journalism, it is not a given that publishers will be able to appropriate more economic value than today. Processes of creative destruction can create temporary or permanent imbalances between the ones who create value and the ones who capture it. Imbalances created by monopolies of the leaders of transformation (from Standard Oil, Google or potentially OpenAI and Anthropic) can become permanent because of scale economies, network effects or switching costs, as observed as early as Schumpeter’s analysis of techno-economic transformations. This might create a need for intervention in the market to ensure socially beneficial outcomes.

Whether or not publishers will be aided by interventions in the market, their economic and journalistic future is likely to depend not only on whether journalism remains valuable. It will depend on whether they can translate their existing capabilities into positions in the value chain where they can capture a portion of the value that is created.

Three emerging positions for news publishers

Figure 2. Publishers positions in the new information value chain

However, new opportunities exist for publishers that can convert their existing capabilities in ways that allow them to control scarce resources in the new value chain. While doing so will not be without challenges, three potential positions are emerging:

News publishers as originators of scarce information (B2A): In this model, publishers continue to provide journalism as real-world events unfold - focusing on the news that does not already exist in the information ecology, and as such, requires human journalistic action and supervision to produce. This journalistic work will likely be increasingly AI-assisted, but focus on information production that requires humans to get physical or emotional access to sources and situations.

News publishers as knowledge providers (B2B): In this model, publishers depart from their traditional focus on producing and publishing news to providing access to their accumulated journalistic knowledge in ways that allow companies and institutions to get answers to the questions they need to perform work. This requires the publisher to build and maintain a knowledge base on top of their existing news production that is structured to make information maintainable (i.e. it must be updated when facts change), verifiable, and retrievable. Direct customers of such knowledge bases are likely to be companies and institutions that need factual knowledge about the domains that the news publisher continuously covers.

News publishers as news experience providers (B2C): In this model, the publisher retains its direct relationship with the broader audience. Instead of exclusively disseminating news stories in articles, videos or podcasts, the publisher provides an intelligent interface through which users can experience and explore news and apply journalistic knowledge to their interests and needs. This makes the news content more personally valuable through interaction, contextualisation, and assistance.

Each position builds on a different journalistic advantage, but each also faces distinct challenges in establishing a defensible market position.

The information originator position is vulnerable to keeping original news content exclusive to ensure bargaining power downstream in the value chain, in competition with other types of information originators and an expanding commodification of news content as AI capabilities grow. To address this competition, publishers must leverage their journalistic capabilities to produce original knowledge that is trustworthy, fresh and relevant to the intentions of individuals, companies and communities while at the same time defending scarcity through control over information access, rights enforcement, uniqueness of reporting, etc. vis-à-vis other information originators. This might also call for publishers to form alliances or consolidate in the originator step of the value chain.

The knowledge provider position will likely face competition from other specialised information services. This must be addressed by protecting the core scarcity – the exclusivity of significant parts of the information in the journalistic knowledge base – through a combination of uniqueness, continuity, and trustworthiness. Also, collecting a broad enough knowledge base might require extending information collection from journalistic reporting to include other complementary types of information relevant to domain specific companies and institutions.

Finally, the news experience provider position is likely to be vulnerable to competition from general AI services (such as ChatGPT or Claude). This competition must be met with a differentiated experience focused on news, building on the publishers’ existing direct relationship with users, which is still strong (and especially so in the Nordic countries). The most powerful differentiation will likely take factors such as editorial judgement, exclusive trustworthy knowledge, and specialised AI-mediated experiences of news content as the point of departure. The specialised AI-mediated news experience can emphasise experiences that prioritises what is deemed editorially important and allows users to discover and explore content in ways that match their intentions and their individual and social contexts. Indeed, a source of valuable differentiation can be tying the personalised news experiences together in an overarching shared experience that lets users know what other users in the community care about and that enables community deliberation.

The vertically integrated knowledge publisher

Figure 3. A vertically integrated news publisher in the AI-driven information ecology.

In this integrated model, the publisher continues to build on traditional journalistic values, methods, and capabilities, although increasingly powered by AI. Its core competitive advantage would lie in its ability to consistently and continuously originate information that AI systems cannot easily (re)produce: uncovering previously unknown facts, first-hand access to sources and events through established networks, independently verifying contested claims. Combined with editorial judgement and accountability, these capabilities will allow publishers to produce original, trustworthy and timely knowledge that is scarce and grows in value in an AI-driven ecology where it can be (re)used for multiple purposes.

Rather than being confined to individual pieces of published content, journalistic knowledge is accumulated in a shared infrastructure, continuously maintained, updated, and potentially enriched with other information relevant to audiences and customers. From there, journalistic knowledge can flow downstream to multiple markets: the news flow licensed to external AI services for grounding and verification of their responses, domain-specific knowledge sold to companies for integration in enterprise AI systems, AI-powered news experiences enabling users to access and explore journalism in ways adapted to individual intentions and contexts.

If adopted, the integrated ABC-model would realise synergies enhancing the publisher’s ability to confront the strategic challenges associated with each of the three positions it integrates. Indeed, the synergies that arise from the economies of scope that the exploitation of information reusability enables might be essential for recovering the still high costs of original reporting (even when empowered by AI), enabling investments in journalism that distributing to one channel alone could not sustain. However, naturally, by entering three markets, publishers will need to carefully manage channel conflicts by traditional means such as differentiating the products, resources and rights offered through each channel.

The future of the journalistic mission

The vertically integrated news publisher is more than a possible economic answer to the transition to the AI-driven information ecology. It is also an answer to how news publishers could reinvent how they realise their editorial mission. By making original and trustworthy knowledge available across a wider range of services and contexts, publishers potentially strengthen their mission in an economically sustainable way.

However, addressing these opportunities does raise questions of how journalistic methods can be reinvented to address these opportunities in ways that remain true to the mission. When information is continuously updated, transformed and reused for different purposes, publishers must develop new methods that ensure the accuracy and provenance of knowledge both at the time of publication and over time. When knowledge is transformed to address individuals’ intentions and contexts, AI-powered news experiences must balance what the user wants and needs, and what is deemed journalistically important. And when journalistic use of AI at scale is likely to become part of news origination itself, developing new ways of ensuring editorial judgement and emotional connection when humans steer processes at higher levels of abstraction will become necessary. Such inventions are likely required to maintain journalism in the AI era, and could also prove to be essential sources of differentiation and competitive advantage in the future information ecology.

The transformation ahead will not be easy. But it offers the opportunity that the destruction of our familiar ways of being a news publisher can be followed by the creation of something new and stronger: A reinvention of the news publisher that strengthens, rather than merely defends, the journalistic mission.