The Scoop
As the sprawling public relations industry scrambles to figure out how to buffer its clients’ brands and reputations through the new medium of artificial intelligence chatbots, some firms have reached a surprising conclusion: The best way to get your client’s message into the output of ChatGPT, Claude, Gemini, and the rest is by talking to journalists.
Firms, whose services now often include regularly testing clients’ reputations with AI models, are finding that authoritative publications — including declining local news outlets and specialist trade journals — shape the results of chatbot queries about a given company far more powerfully than a social media campaign or Reddit thread could. The result is a striking reversal of the status quo at a moment when PR executives had begun to enjoy the social media-era option of ignoring journalists entirely.
“Earned media still matters, but not the way people think,” said Carreen Winters, who leads the reputation practice at MikeWorldWide, using the trade term for independent reporting.
The firm is launching a service this week called “PreBunk” that’s designed, according to a draft press release shared with Semafor, to provide an “ongoing proactive ‘education’ of the LLMs about your company and its reputation.”
Consumers, according to Winters, say, “I’m not going to trust earned media — I’m going to trust the internet.” But these LLMs’ sources lead back to journalism, something she said can sometimes be a hard sell to executives who thought they no longer had to deal with pesky reporters.
“Sometimes it’s a small trade publication that your client has said, ‘Nobody reads that anymore,‘” she said. “Sometimes it’s a hometown newspaper.”
Other firms are reaching similar conclusions. “Earned media and owned content [that is, pages on a company’s own website] are the primary drivers of how GenAI platforms recommend and describe brands and products. It’s not even close,” said Brian Buchwald, who leads Edelman’s global product, data and AI strategy. He said the firm carefully tracks the sources of LLM answers, which vary widely based on industry and brand. LLMs’ assessments of an enterprise tech company’s reputation, for instance, drew from Wall Street Journal coverage and research reports from Gartner.
“You can make a big difference very quickly with the right content and campaign choices and who writes about it,” he said.
Rand Fishkin, the founder of the audience research firm SparkToro, wrote last year that, for instance, LLMs appear to rely heavily on professional review sites like Eater when recommending restaurants. For brands looking to stand out, “that’s gonna be a PR process and a pitch process, but is it worthwhile? Absolutely,” he wrote.
He recently headlined another post: “Unpopular Opinion: Public Relations is the Future of Marketing.”
In this article:
Know More
The PR industry is navigating the rise of AI in parallel with the overlapping but more technical SEO trade, which is adjusting its sights from bringing clients’ websites up search results to elevating them in AI excerpts on Google and elsewhere. One place they converge is in encouraging companies to add pages to their website aimed at LLM, not human, consumers.
The SEO professionals are finding, however, that for now AI is largely relying on the same rankings that search engines use — though sometimes in unpredictable ways. A brief from the enterprise SEO marketing company BrightEdge, for instance, cites as “one of the most important discoveries” about Anthropic’s Claude the fact that it relies on the lesser-used search engine Brave for its rankings; companies will need to ensure they’re being indexed by Brave to feed their official line to Claude. Another brief wrestles with the subtle differences between Google’s AI Overviews and more traditional search rankings, with the LLMs answering “anticipated questions,” not just the ones consumers are asking.
The SEO field has long been engaged in a game of cat-and-mouse with Google and other search products, and spokespeople for LLM companies didn’t respond to inquiries about how they view these efforts.
Ben’s view
There’s something heartening, from the perspective of the humans in the media business, about the practice of gaming digital media becoming less technical, after a long march in which advertising and marketing were essentially swallowed by adtech and practices like SEO.
Dealing with LLMs is “more like traditional PR than it is like SEO,” Ben Worthen, a former Wall Street Journal reporter who founded the agency Message Lab, told me.
That’s good news, in particular, for the PR industry, which gainfully employs its share of human beings, as well as some former journalists.
But even if the LLMs find this kind of authoritative journalism valuable, and even if companies will pay to employ publicists to pitch their stories, it’s not clear where that process meets news organizations’ business models.
For instance: What, exactly, is a trade publication that offers valuable and authoritative service to LLMs even as humans stop reading it? A research service for AI? If the handful of firms training and maintaining LLMs really think that the authoritative reporting on small industries or local areas is valuable, they may have to pay for it — becuase nobody else seems to want to.
Room for Disagreement
The darkest warnings about the power of AI have to do with the “liar’s dividend” that renders accurate journalism pointless. The theory is not that deepfakes will persuade people to believe anything in particular, but that they’ll make people disbelieve everything. Bobby Chesney and Danielle Citron coined the term in a 2019 essay arguing that “a skeptical public will be primed to doubt the authenticity of real audio and video evidence. This skepticism can be invoked just as well against authentic as against adulterated content.”
Notable
- AI relies on original journalism, a Brookings Institution report argued: “Without access to human-created, high-quality content that is a relatively accurate portrayal of reality — and that journalism provides — the foundational models that fuel machine learning and generative AI applications of all types will malfunction, degrade, and potentially even collapse, putting the entire system at risk.”
- Or, as Joshua Rothman put it in a relatively optimistic New Yorker essay, “A.I. could improve the news — if it doesn’t destroy it in the process.”
- The AI effect on search is here, as searches in Safari recently fell for the first time ever, according to an Apple executive.