In an era defined by rapid and seemingly mindboggling technological advancement, artificial intelligence (AI) is reshaping nearly every aspect of knowledge creation and dissemination. Nowhere is this transformation more disruptive - or potentially more empowering - than in the domain of scientific, technical, and medical (STM) publishing. As AI-driven tools such as Google's Gemini integrate directly into search engines, enabling users to access concise, context-aware answers without ever clicking through to original sources, traditional traffic patterns and user engagement models are being upended.

The Rise of AI Overviews and the Erosion of Organic Traffic

One of the most immediate and tangible impacts for STM publishers is the decline in web traffic. With Google's AI Overviews offering instant summaries drawn from indexed content, up to 20 to 60 percent of organic traffic could disappear, according to recent projections. Where once a user might visit multiple journal pages or dig through a publisher’s archive for insights, they are now increasingly served AI-curated syntheses directly on the search results page. The implications for content discovery, engagement, and monetization are significant.

This shift places publishers in a difficult position. Publishers still produce the authoritative, peer-reviewed content that AI models rely on, yet they are no longer guaranteed the user interactions that support their data strategies and revenue models. This breaks-down the researcher to publisher relationship. The implications of this are potentially far worse than revenue decline; it could erode the virtuous circle that exists across the ecosystem, including  institutions, researchers/authors, funding bodies and the publishers.     

The Dual Threat: AI and the Demise of Cookie-Based Tracking

At the same time, the phase-out of third-party cookies is making it harder for publishers to understand and reach their audiences. With 30 to 50 percent of web traffic becoming effectively invisible, STM Publishers are losing the analytics and targeting tools that once allowed them to build user profiles, personalize content, and run successful outreach campaigns. 

This combination of AI-driven search disruption and limited visibility through data-tracking creates a serious need for STM publishers to rethink how they create and deliver value for the readers.

Communities as the Countermeasure to AI Disintermediation

Instead of competing directly with AI, leading STM publishers are turning to one of their most underutilized assets: the scientific community itself. Platforms like Zapnito promote the idea of transforming passive readers into active participants by building interactive, knowledge-sharing communities.

By creating digital environments where researchers can collaborate, connect, and share insights asynchronously, publishers can offer something AI cannot. Peer-to-peer interaction, live discussions, tailored content hubs, and expert Q&A sessions foster user-generated content that is rich in context and deeply human.

According to HubSpot and McKinsey, audiences trust peer-generated insights more than algorithmic outputs. Communities also encourage brand loyalty and drive return visits more effectively than traditional content delivery models.

Personalization Over Aggregation: A New Model for Engagement

AI tools work by aggregating and summarizing vast amounts of data. Community platforms, in contrast, provide more personalized experiences that reflect individual user needs, roles, and interests. 

And this personalization is not algorithmically decided, it is decided by the humans that follow each other.

Custom groups, topic-based channels, behavior-based notifications, and curated content collections help users navigate in a way that feels more relevant and intuitive.

In this model, publishers regain ownership of the user experience and their data. With first-party data collected through registrations, interactions, and engagement patterns, they are better equipped to optimize offerings, track behavior, and build lasting relationships.

The Value of What AI Cannot Do

There are essential aspects of scientific exchange that AI simply can’t replicate. Collaboration, mentorship, peer validation, and live events require human connection, emotional nuance, and shared context. These are foundational elements of scholarly communication and remain out of reach for even the most advanced AI models.

By creating spaces that enable these types of interactions, publishers can protect against the dehumanizing effects of automation and also improve the depth and quality of scientific communication.

And at the same time, develop a dynamic relationship with the Researcher.

AI Summaries Cause Sharp Drop in Audience Traffic

Recent studies show that AI-generated summaries in search results are significantly reducing website traffic. Analytics firm Authoritas found that pages previously ranking first on Google can lose up to 79% of their traffic when placed below an AI overview. Users are often getting the information they need from summaries alone, without clicking through to the original content.

UK publishers like MailOnline report drops in clickthrough rates of over 50% on both desktop and mobile. A Pew Research Center study also found that users click a link beneath an AI summary only once in every 100 searches.

Industry groups have raised concerns with regulators, arguing that Google’s AI features threaten the visibility and sustainability of quality content. While Google contests the findings, publishers say they lack the data to assess the true impact - calling for more transparency and urgent regulatory scrutiny.

A Strategic Imperative for STM Publishers

STM publishers face a critical decision point. One option is to rely on distribution through third-party algorithms that prioritize convenience over credibility. The other is to invest in strategies focused on community building, collaboration, and direct engagement with their audience.

The case for online communities is stronger than ever. It helps drive loyalty and repeat visits, provides valuable insights, and supports sustainable engagement over time. Most importantly, it offers a way to retain relevance and create more value in a landscape where AI threatens to reduce even the most authoritative content to a brief summary or question and answer experience.

To succeed, publishers must act quickly. This means building the right infrastructure, rethinking how content is distributed, and focusing on first-party relationships across their ecosystems. By doing so, they won’t just stay relevant, they will strengthen their role in advancing science through meaningful, human-centered connections.