How PR agencies are using AI

Brand and corporate PR—which has historically relied on interpersonal communications—is increasingly being influenced by artificial intelligence as public relations agencies adopt AI tools to monitor and place stories about their clients. In some cases, AI is used to boost the odds of getting favorable coverage. 

The use of AI in PR has been underway for years but has taken on new relevance as AI becomes more sophisticated and agencies are tasked to bolster the public reputation of brands.

PR giant Edelman—which for some 20 years has published an annual “Trust Barometer” that tracks factors influencing how corporations are viewed—has begun integrating AI into the offering. The “Trust Management Platform,” which was publicly launched this January, uses predictive and analytical technology that is designed to show clients through charts, maps and scores which consumers trust them, how much, and when this trust fluctuates, according to a demonstration viewed by Ad Age.

The product joins other AI-informed offerings deployed by a range of PR shops that are meant to pinpoint relevant journalists, generate pitches and address public sentiment more quickly. 

“It’s (become) all about how you actually make the insights usable and practical,” said Amber Boyes, a director analyst at Gartner. Though the industry has used AI for a while for tasks such as monitoring media, “the buzz around ChatGPT and generative AI more generally presents new use cases,” she said.

The PR industry’s use of AI comes as creative agencies deploy it in varied ways, including using image generators such as DALL-E 2 in ad campaigns.

AI is “no longer seen as giving a competitive advantage, it’s about ensuring [companies are] not left behind,” said Noshir Contractor, the Jane S. & William J. White professor of behavioral sciences in Northwestern’s McCormick School of Engineering. 

Along with Edelman, PR agencies and holding companies including Stagwell, Interpublic Group of Cos. agency Weber Shandwick and marketing and communications agency Boathouse have created tools to measure public sentiment or target relevant journalists.

Edelman frames its trust data as what sets the platform apart. “This is a problem that only Edelman could have solved, and a tool only Edelman could have built,” Global Chief Digital Officer Tristan Roy stated in an email. “No one else has this repository of data, or this analytical and now technical understanding of the drivers and detractors of trust.” Trust is defined as the long-term beliefs a person has about a company, according to an Edelman spokesperson.

Edelman demonstrated the platform to Ad Age under the condition that brand names remain anonymous.

The dashboard of a major streaming content platform, for example, showed that its trust dipped after the brand ran content that was deemed transphobic. Clicking on the dip in the chart brought up a list of articles ranked by the effect Edelman said they had on trust. Brands can identify which journalist wrote the article. In a separate part of the dashboard, brands can also see which audiences trust them the most, including by region of the country. For instance, users can examine a map of the U.S. with audience breakdowns by trust level, or by race and age—such as what its “trust score” is with southern Gen Z consumers.

Edelman began offering the platform to clients last October, onboarding 60 brands before making it more widely available in January. The agency declined to specify how many clients are currently using the dashboard.

Other tools include Stagwell product PRophet, which was created in 2020 and is part of the holding company’s suite of marketing tech tools. The platform uses machine learning, natural language processing and semantic search to identify journalists most likely to be interested in a pitch, said Aaron Kwittken, its founder and CEO. The technology also predicts whether the resulting article will skew positive, negative or neutral by giving each journalist a color—green, red or yellow, respectively.  

The platform, which has approximately 500 users annually according to Kwittken, was created to address the “outdated, analog system” of emailing long lists of journalists provided by third parties such as Muck Rack which results in “very frustrated journalists, wasted time, mismanaged expectations and poor results,” said Kwittken.

The product will incorporate generative AI into its services by the end of the first quarter to allow users to generate pitches and then refine pitches, said Kwittken. 

Independent agency Boathouse created its first AI-powered tool in January. The “Narrative Transformation” tool, which uses technology from software company NetBaseQuid and media monitoring company Signal AI, gathers data from websites such as Reddit and Glassdoor to understand public sentiment through trending hashtags or employee attitudes.

Adding to the noise

But the tools are not foolproof. If the product or service that’s being pitched is weak, the targeted pitch will still be overlooked and “just add to the noise,” said Lindsay Grace, Knight chair in interactive media and associate professor at the University of Miami School of Communication. He also suggested that over time, the tools will become less effective as journalists and other recipients of the targeted outreach learn to identify AI-generated pitches, which in some cases can be screened with the help of technology.

Boathouse emphasizes the continued need for “human intelligence to augment the artificial intelligence,” President of Boathouse West Coast Peter Prodromou stated in an email. 

Weber Shandwick has used AI for several years, according to a spokesperson. The PR firm’s Media Security Center, launched in 2021 with technology from risk intelligence platform Blackbird.AI, focuses on disinformation by identifying public sentiment and provides training sessions for executives and teams on misinformation. 

Weber Shandwick also has an Analytics+Intelligence team, created in 2020 and previously called its Global Intelligence division, that uses Weber Shandwick’s own AI models to understand how consumers feel about a brand, according to a spokesperson.

Omnicom-owned FleishmanHillard has “used AI in monitoring, analytics and intelligence for some time,” Global Platforms Lead Jared Carneson stated in an email. The company is listed on the testimonials page for Signal AI along with firms such as BCW, Boathouse and Next PR. Signal AI was unable to be reached for comment.

FleishmanHillard has “experimented with chatbots in customer service, logo/image recognition, language recognition and generation, AI-driven audio modification, AI-centric automated creative workflows and more in addition to the generative AI space too,” said Carneson. 

“We are however cautious in the true integration of AI of a generative nature into ongoing work until issues of accuracy, bias and content custody are addressed,” he said. FleishmanHillard currently uses generative AI for inspiration rather than work and will do so “until policies and standards for use are clear,” Carneson said.

Questions remain about the technology’s fallibility–Google’s Bard chatbot recently gave a wrong answer in its first demo. There are also uncertainties about ingrained biases, said Jacob Appel, an algorithmic auditor with auditing company ORCAA. 

“If the PR industry is worried about stereotypes or saying the wrong thing, AI-generated (content) would incorporate stereotypes we encounter all the time,” Appel said. For example, when writing a pitch, a chatbot could default to using male pronouns for a creative director, simply because there are more examples of creative directors being men due to historic bias, he said.