The traditional landscape of public relations, once defined by the linear relationship between practitioners and journalists, is undergoing a profound transformation as artificial intelligence redefines the mechanisms of information discovery and influence. During a recent session of "AI Pulse," a monthly briefing hosted by the Public Relations Society of America (PRSA), industry experts convened to discuss how the rise of generative AI search engines is fundamentally altering the strategic mandate for communications professionals. The consensus among the panel was clear: while the core tenets of storytelling remain, the target of those stories has expanded from human audiences and search engine algorithms to include the complex Large Language Models (LLMs) that now serve as the primary interface for millions of consumers seeking information.
Angela Dwyer, APR, Vice President of Insights at Fullintel, emphasized that artificial intelligence is not merely a tool for internal efficiency but is actively "changing how we create content." She noted that "AI search is a huge channel, in terms of influence," suggesting that for the modern PR professional, securing a placement in a high-tier news outlet is no longer the final step in a successful campaign. Instead, that placement must now serve as a data point—a "signal of authority"—that AI models can crawl, synthesize, and cite when answering user queries.
The Enduring Primacy of Earned Media
Despite the technological shift, the value of traditional earned media remains remarkably high, albeit for new reasons. Research conducted by Muck Rack, which was highlighted during the briefing, reveals a significant trend in how AI models select their sources. According to the data, AI models demonstrate a clear preference for credible journalism over paid or promotional content. Approximately 94% of the sources cited by leading AI models are non-paid content, and of those, 84% are earned media mentions—the specific type of coverage that PR professionals work to secure through media relations.
This preference for earned media is viewed as a validation of the PR industry’s historical focus on third-party credibility. Martin, the Managing Director of Media and Influence for Allison Worldwide in New York, described this trend as "super exciting for comms professionals today." He argued that the findings create a mandate for the industry to analyze how AI systems interpret the work of publicists. AI "answer engines," such as Perplexity or the AI Overviews in Google Search, often utilize unpaid, third-party content as a primary signal of authority. When an AI model evaluates information to serve to a user, it prioritizes sources that have undergone the gatekeeping process of professional journalism, viewing them as more reliable than corporate press releases or paid advertisements.
A Chronology of Search and Discovery
To understand the current shift, it is necessary to view the evolution of digital discovery over the last two decades. In the early 2000s, discovery was dominated by keyword-based Search Engine Optimization (SEO), where the goal was to rank on the first page of Google. By the 2010s, this evolved into social media discovery, where algorithms prioritized engagement and "virality."
The current era, beginning roughly with the public release of ChatGPT in late 2022, marks the transition to Generative Engine Optimization (GEO). In this new environment, the goal is not necessarily to drive a "click" to a website, but to be the "citation" that the AI provides as the basis for its answer. This represents a move from a "link economy" to a "citation economy." As Dwyer noted, while website traffic from traditional search may be declining, the influence of the content remains high because users are simply consuming that information through a different medium—the AI-generated summary.
Model-Specific Sourcing Patterns
One of the most critical insights from the PRSA briefing was the revelation that not all AI models are created equal in their sourcing habits. The research indicates that different LLMs prioritize different types of media based on their underlying architecture and corporate affiliations:
- ChatGPT (OpenAI): This model remains the most reliant on traditional earned media, with news articles and journalistic features making up approximately 35% of its cited answers. For PR pros, this means that a classic media relations strategy remains highly effective for influencing ChatGPT’s output.
- Claude (Anthropic): Positioned as a more scholarly and safety-conscious model, Claude tends to favor research papers, government white papers, and academic sources. Earned media falls to third place in its hierarchy of credibility, suggesting that for industries like biotechnology or public policy, a focus on "owned" technical content is more effective.
- Gemini (Google): As expected, Gemini leans heavily into the Google ecosystem. It frequently cites content from YouTube and other Google-indexed properties. In this environment, earned media is often secondary to video content and direct data from Google’s massive web index.
- Copilot (Microsoft): Similar to Gemini, Copilot prioritizes its parent company’s assets, specifically LinkedIn. It also relies heavily on the Bing search engine index. This makes executive thought leadership on LinkedIn a primary lever for influencing Copilot’s responses.
The Rise of Discovery-Based Communities
Beyond traditional news outlets, the panel identified a growing trend toward "discovery-based communities" as vital sources for AI training and real-time retrieval. Martin noted that platforms such as Reddit and Quora are becoming essential pillars of earned visibility. Because AI models are trained to mimic human conversation and problem-solving, they often look to these forums to understand how real people discuss products, services, and trends.
Furthermore, LinkedIn and YouTube have emerged as critical platforms for "earned influence." Martin told Dawn Robinette, APR, Fellow PRSA—who moderated the session for regular host Ray Day—that executive thought leadership on LinkedIn is no longer just about networking; it is about providing the AI with high-authority, person-centric data points. When an executive publishes a deeply researched article on LinkedIn, it increases the likelihood that their perspective will be included in AI-generated summaries regarding their industry.
Strategic Shifts: From Promotion to Education
The shift in how information is consumed necessitates a change in how content is written. Angela Dwyer’s research indicates that "educational content" is the most frequently cited type of material across all major AI models. This is closely followed by "answer articles"—content specifically designed to provide a direct, factual response to a common question.
"People don’t like promotional content," Dwyer explained. "But we do like our questions to get answered." This distinction is vital for PR strategy. A press release that focuses solely on the "innovative" and "market-leading" qualities of a product is likely to be ignored by an AI model looking for objective facts. Conversely, a white paper or a contributed article that explains how a specific technology works or why a certain market trend is occurring provides the "educational" value that AI models prioritize.
In certain sectors, such as healthcare, "owned" content—the information found directly on a company’s website—is still viewed as highly credible by AI models, provided it is presented in an authoritative, non-promotional manner. This suggests that the future of PR lies in a hybrid approach: maintaining strong media relations to secure third-party "authority signals" while simultaneously developing a robust library of educational owned content.
Implications for the PR Profession
The implications of these findings are far-reaching. First, the metrics of success in PR are evolving. While "impressions" and "unique visitors per month" (UVM) have long been the industry standard, practitioners may soon need to track "citation share"—the frequency with which a brand or client is cited by major AI models in response to industry-relevant queries.
Second, the role of the PR professional is becoming more integrated with technical SEO and data science. Understanding the "crawlability" of a news site or the "authority score" of a LinkedIn profile is now as important as knowing how to write a compelling pitch.
Finally, the discussion underscored that the human element of PR—building relationships with journalists and influencers—is actually becoming more important, not less. As the internet becomes flooded with AI-generated "slop" or low-quality synthetic content, the value of a verified, human-written article in a respected publication increases. AI models are programmed to seek out this "human-in-the-loop" content to ensure their own outputs remain accurate and trustworthy.
The "AI Pulse" briefing served as a reminder that while the tools of discovery are changing, the fundamental need for trust and authority has not. For PR professionals, the challenge is to navigate this new digital landscape by ensuring their clients’ messages are not just heard by humans, but are also "read" and "trusted" by the machines that increasingly guide human decision-making. By focusing on educational content, leveraging discovery-based communities, and maintaining a rigorous focus on earned media, the communications industry can ensure its continued relevance in the age of artificial intelligence.
