Artificial intelligence has promised to revolutionize our lives, taking over the mundane tasks of daily existence, from prewriting “smart” email replies to driving our car through rush hour traffic.
In the PR realm, AI has been touted as equal parts something to celebrate (no more manual coverage reports!) and fear (er, so long, means of employment). But the truth, as usual, lies somewhere in between.
Some form of intelligent technology is already embedded in the PR industry, from the tools we use to find new audiences and monitor evolving conversations to modern media placement. Bloomberg News uses AI to generate coverage on some 3,500 earnings reports every quarter. The Associated Press has followed suit, using AI to create articles on minor league and college sports in addition to company earnings.
But too often when PR folks talk about AI (which doesn’t happen often enough), the conversation hedges around job losses and deeper insights into information we already gather—like audience reach and sentiment—and not how we can revolutionize our roles in building and promoting brands.
There’s merit to fine tuning what we already do, but PR pros both in-house and at agencies are getting lapped by their marketing counterparts who have completely embraced technological innovation to personalize and automate what messages their customers receive.
Just because PR trades in relationships and storytelling doesn’t mean that technology won’t come for us too. Algorithms already connect us to new social circles and help target messages and topics that will motivate desirable audiences. But media databases and social listening tools are only the beginning. If we’re going to modernize PR, we should start by correcting these AI wrongs:
Most AI-branded solutions don’t have much, if any, AI under the hood.
There’s nothing “artificial” about the intelligence powering the most common PR tools. We (or our tech vendors) spend significant amounts of time teaching machines how we want them to think, then fixing the algorithm’s mistakes before we put that information into action. If you’ve ever spent time retagging sentiment on a media report, you know the limits of computer intelligence applied to complex human emotions.
Perfect machine learning isn’t required for true AI but it is an important indicator that AI is about to take over a specific task. Only about 12% of PR tools and processes utilize AI at a significant level, according to a 2018 study by the Chartered Institute for Public Relations, the largest membership organization for PR practitioners outside of North America.
PR pros would be well served to master AI in the areas where it threatens to take over from human effort, particularly data management and analytics. But we must also get more tech savvy overall.
We can’t just be focused on reaching audiences where they are. We must start thinking about how we can reach more people with personalized messages in less time than it takes to secure traditional mass media. We need to show models for how different messages could reach and influence audiences. And we need to identify potential online threats and actively monitor for precursors, effectively negating crises before they start.
Otherwise, we’re only chasing consumer behaviors and attitudes, not getting in front of them. Which leads me to another of PR’s AI problems…
We’re not using AI to predict or alter behaviors.
While PR people are focused on capturing and reporting the impact of our past work, our marketing counterparts have moved on to predicting behavior with deep learning, a buzzy phrase that means using computing power and artificial intelligence to find the patterns in vast sets of data.
Humans are complicated, novel creatures. We don’t like to think that our behaviors can be predicted, and hey, we still have free will. But while marketers dream of a future where they can pinpoint new clients based on their recent behaviors, PR pros too often stop at the press hit or social media mention and don’t ask what happens next (beyond what a recent win means for the next chapter of their campaign).
AI can already tell you what content people will respond to based on what they’ve read or shared before but too few publicists take this intelligence from the realm of owned media and apply it to earned or shared opportunities. In many cases, the reason for this is simply that the data doesn’t belong to us. But just as journalists are increasingly comfortable being asked to include a backlink in their coverage, they may also be willing to shed some light on their highest-traffic articles (or you could always do a little sleuthing yourself) and what people also read.
PR folks can no longer run from probability and statistics. We must get cozy with the data we do have and get out of the mindset that we aren’t in control of driving people to action. Increasingly, that is the job we’re being asked to do.
If nothing else, we need to get serious about the task of data management. Because those practitioners who don’t have data sets to use when deep learning takes off will be caught flat footed—and they may never catch up.
We’re not flagging fakes.
What do Nancy Pelosi and Donald Trump have in common? It’s not a joke. They’ve both had their public images altered in “deepfake” videos and they’re far from the only ones.
Yes, fake news has broken into the visual realm. While you don’t need advanced AI to create a video of something that never actually happened, deepfakes build on deep leaning to make frame-by-frame editing an elementary process. Unfortunately, the technology that detects such manipulations isn’t so intuitive and by its very nature, is always one step behind.
Better fakes and more-convincing misinformation are coming, just in time for the 2020 election. But politicians aren’t the only ones in the crosshairs. Companies and individuals are also at reputational risk, with social media and traditional news outlets reluctant to shut down the deepfakes entirely, since not all fake videos are nefarious. It all adds up to one enormous PR problem.
Other than investing in AI that detects altered videos as it becomes available, PR pros must start to think about where they’re vulnerable and create plans to address those weaknesses. That may mean watermarking earned content both visually and digitally, or constructing response plans for when a deepfake hits.
The reality is that AI is coming and once it reaches critical mass in the PR industry, it will be too late to get up to speed. Correcting these top wrongs may not be enough to keep the pace in the next 3- to 5-years but they are an important place to start.