The USC Annenberg center for public relations publishes every January its “Relevance report” and this year’s edition is mostly about AI. It is more an anthology of views from 50 corporate folks, some in PR and some in other industries that are PR-adjacent. Even if you aren’t a PR pro or a journalist, the 111 page report is worth a download and at least an hour of your time.
Here are some of their insights that I found most, uh, relevant.
The lead off piece is by Gerry Tschopp, head of Experian’s comms team. They have been using Chat GPT to speed up their responses with stockholder communications, social media analytics. “What once required hours of research and revisions is now handled in a matter of minutes” with the AI chat tools.
Many of the authors point to how AI can automate the mundane, everyday tasks such as organizing databases or formatting reports or providing other suggestions to improve the quality of first drafts. Jaimie McLaughlin, a headhunter, uses AI to enhance candidate matching for recruitment purposes. Pinterest is using AI to reorder its content feed to focus on inspirational and more positive and actionable content. Grubhub is using AI to design new ad campaigns that focus on more emotionally-charged moments, such as the changes wrought with a newborn, or creating a first draft of a press release in a matter of seconds. Microsoft (who as a corporate sponsor has several contributions) has redesigned its transcription workflow of interviews using AI, as shown here. And Edelman PR is using AI to be more proactive at client reputation management and in improving trust on specific business outcomes. This was echoed by another PR pro that went into specifics, such as using AI to detect and analyze situations that could turn into a full-blown crisis by automating data collection in real-time, tracking the evolution of any issues as they unfold. AI can do sentiment analysis from this data, something that used to be fairly tedious manual work.
ABC News is using AI to debunk AI-generated viral videos, because they are so easily created. As one producer put it, “Here’s what keeps me up at night: It takes eight minutes and a few dollars to create a deepfake. Truth, measured in pixels and seconds, has never been more fragile.”
It is clear from these and other examples peppered throughout the report that, as Gary Brotman of Secondmind says, “AI tools have become integral to everything from automating social media monitoring and trend analysis to enhancing campaign measurement. ChatGPT has become my co-author for just about everything.” His essay contains some interesting predictions of where AI is going over the next five years, such as with hyper-personalized communications, predictive content creation and eroding knowledge silos everywhere. Yet despite these innovations, he feels that AI adoption has been slower and less impactful than many predicted because we have neglected the human element.
“The integration of AI into PR isn’t a short-term project with a finite end date — it’s an ongoing journey of innovation and refinement,” says one AI executive. And I think that is a good thing, because AI will bring out the lifelong learners to experiment and use it more. It will encourage us to think beyond the obvious, to find interesting connections in our experiences and contacts.
And there are plenty of tools to use, of course. Dataminr (newsroom workflow), Zignal Labs (real-time intel), Axios HQ (writing assistant), Glean (various AI automated assistants), and Otter.ai (transcriptions) were all mentioned in the report. I am sure there are dozens more.
In a survey conducted by Waggener Edstrom PR, the top four concerns about adopting AI tools for PR purposes included information security, factual errors and data privacy, all mentioned by almost half the respondents. That seems about right.
Sona Iliffe-Moon, the chief communications officer at Yahoo, sums things up nicely: We have to focus on the communications that matter most, use AI for scale not strategy, and put authenticity and trust but verify with humans. Trust but verify — now where did we hear those words before? Luckily, we have chatbots and Wikipedia to help out.
Interesting lead in
As a PR pro my my biggest concern is GIGO. A language model “writing” a release, pitch, or any other content is scraping all similar examples and spitting out a piece that resembles them. Cool, but 90% of press releases are garbage, and that’s being generous.
I remember an AI-written news release that a colleague sent me, ostensibly to help me get a first draft started. It was quite possibly the worst thing I’ve ever read. It hit every “don’t” on the bingo card: “[Company] is excited to announce…” and so forth. College interns could produce better copy on their first try.
GIGO applies to analytics, monitoring tone or reputation, and others. Bad data = bad results.
Admittedly I don’t work for a huge firm that needs to think about this junk. Phew.