When Not to Use Artificial Intelligence in Digital PR

Edited by Emilie Martin

In recent years, artificial intelligence (AI) has been catapulted into the mainstream. No longer a confusing tool hidden behind the computers of tech nerds, we’ve seen it talked about by every news publication imaginable with endless Twitter (X?) threads giving tips on how to utilize it.

Hell, your mom has probably brought it up by this point.

From an industry point of view, AI has divided marketers of all kinds. Will it replace your job? Does it do your work better than you? Maybe you’ve even seen companies liquidate, citing AI as their downfall.

No matter your point of view, AI is a valuable tool for almost anyone. In SEO, we’ve seen it used to create content at a huge scale, identify target keywords in record time, and even create images from scratch for those of us who don’t have a designer bone in our body.

Love it or hate it, these tools have become second nature for all kinds of marketers

One of the most popular ones being OpenAI’s ChatGPT, which will be used throughout this article. The recent leadership struggle at OpenAI has dominated social media discourse recently, not just within the AI niche but in general. Let’s see what it’s all about.

Are you leveraging AI's LLMs tools (like chatbots such as ChatGPT, Bing Chatbot, Chatsonic, etc.) or functionalities (those integrated with existing tools to generate ideas, content, etc. like Frase) to support your SEO tasks or efforts? 👀🤖

— Aleyda Solis 🕊️ (@aleyda) May 9, 2023

In some cases, AI can be a useful tool in the digital PRs toolbox, namely when working with large amounts of data. AI is able to digest research to pull out interesting stories, find sources for your campaigns, and even suggest survey questions for interviews.

However, it’s starting to pick up a bad name for itself due to those who use it where it isn’t best placed.

AI doesn’t necessarily prioritize empathy and fostering relationships with humans (except for maybe the one that professed its love to one user). It’s much more pragmatic than that. It’s artificial, making its communication artificial, and as humans, we can always tell if we’re communicating with one of our own.

As someone who is quite active in the digital PR community, founding a search-first PR consultancy, and editing a content campaign newsletter, I’ve seen my fair share of how AI has been used in our industry, but not always in a positive way.

So, although AI has helped us speed up processes and given us a whole new array of tools to wield at our jobs, when should we be more cautious about working with it?

One of the areas where people often note AI falls down is creativity.

At agencies, the ideation phase of a campaign is already generally the most starved. With a lack of time and support, we often cut corners to produce something that has already been done. With the introduction of AI, the problem only perpetuates.

If you use AI to cut corners and speed up your process, it’s unlikely that you’re going to come up with an idea that’s going to cut through journalists’ already overflowing inboxes.

The key to using AI in ideation is to break the process down as small as possible and only use AI in select areas, not as a whole.

Come up with concepts (from scratch)

When it comes to generating an idea from scratch, at face value, AI might seem like it has done a pretty decent job. Though soon, you'll find that it often provides the same idea over and over again and is rarely specific to your business.

Asking for a 'digital PR campaign' will often have the tool come back with traditional ideas such as interviews, case studies, and collaborations. While these can work well in practice, many need an extra step to become truly newsworthy.

Below, I asked ChatGPT to come up with five ideas around solo travel. Anyone with an eye for digital PR will realize quickly that these campaigns lack substance and would be risky in their raw form.

At one point, it even suggested a contest. Oh no.

I even tried to take it a step further and pull out specific data points for some of the suggested ideas, but the tool tried to point me to research articles to find trusted data sources myself. Not so helpful.

Brand nuance and emotional intelligence

Another struggle I encountered when using AI for ideation was its complete lack of emotional intelligence.

Take the below as an example. Although the prompt isn’t the most specific in the world, ChatGPT seems to have understood what we’re looking for. I asked it to come up with a campaign idea for a fast fashion brand, which it did in record time:

As you can see, the first idea it defaults to is looking at sustainability in fast fashion. The AI is right in thinking this is an important issue covered extensively by the media, but it fails to understand that it might be a pretty risky topic to explore.

It bypasses any sense checks that we as humans would be able to pick up on. From a PR perspective, we know that talking about sustainability as a fast fashion brand, especially with no initiatives in place as a company, opens us up to an absolute PR nightmare.

It’s also worth mentioning that the idea isn’t something that is particularly easy to execute but would realistically need a complete overhaul of the business model. AI doesn’t understand budgets, timings, how realistic it is for a specific business to implement campaigns, and many more nuanced factors.

Maybe adding more context to the prompt would help this, but even then, the idea at its core is inappropriate at best.

Brainstorm ways to amplify the idea

To give AI some credit, where ChatGPT can be slightly more helpful is in providing campaign elements and touchpoints to ensure it works over multiple platforms.

I say this with a huge caveat. If you've never tried to amplify a campaign concept before, it can act as a good starting point or sounding board. Otherwise, we hit some issues very quickly.

The more you use the tool, the more you’ll notice that these ideas are often regurgitated. Every campaign has an influencer tie-in, all digital PR projects have a website ‘hub’, and after more than a few tries, you’ll find yourself pitching a branded hashtag for every campaign.

Some of these will be helpful and give your campaign extra arms to exceed over platforms, but mostly, they aren't. I even tried to add more context to the prompts to create campaign elements that would help win a Cannes Lion award. This took things in a different direction, but still, ChatGPT's creativity appeared to stop at the fifth suggestion to 'launch a podcast.'

AI can be incredibly helpful when producing a campaign. From text-based AI tools like ChatGPT to image-generation tools like Midjourney, there’s so much at our fingertips to use.

However, campaigns should always be ethical, look sleek, and include robust data. While AI can produce most content at the touch of a mouse, not all of it is appropriate for consumption by your customers. Creating sub-par content can harm brand perception and even scare away customers.

Writing landing pages for PR campaigns

If you throw your idea into an AI tool and ask it to come up with 2,000 words on the campaign findings, within minutes, you'll be given landing page content you can throw up on the website. However, it’s going to be severely mediocre at best.

The point of a landing page in PR is to tell a story, whether that’s with data, a stunt, or something else that gives journalists and readers extra information on the topic. AI will likely miss out on context and information that is important for the reader to understand the study or campaign. No one knows more about the campaign than those creating it.

While it’s generally understood that Google doesn’t penalize AI content (yet), if you create a blog that’s almost entirely made from AI content with no human intervention, your readers will likely find the content jarring. This will eventually lead to your client's reputation being affected and potentially people seeking out alternative content that reads more naturally.

Short-form copywriting

While AI works best with longer forms of text, when it’s challenged to write short-form copy, it doesn’t quite understand the task. Of all the content you can produce, the short-form copy used for things like headings, billboards, and within advertising are some of the most human examples of text you can find. It’s often smart, multi-layered, and punny.

Give ChatGPT a prompt to have a bit more fun with the copy, and it comes up with the blandest dad jokes you could ever imagine.

In early 2023, ASDA released print ads about how they were experiencing a food shortage with copy like “Romaine calm.’

This is how ChatGPT handled a similar brief:

It’s the ‘Salad SOS!’ for me.

Provide up-to-date and ethical data

If you ask AI to provide you with data, it’s usually happy to oblige. However, without checking the information it gives you, you’re likely to end up with data that is out of date or incorrect.

The most free version of ChatGPT specifically only has data up to January 2022, so any information you do ask it for will likely not be usable. I asked the tool to give me the ten most populated cities in the world. It provided me with the information I needed, albeit with a disclaimer:

I did my due diligence and checked this against official data from World Population Review, and of course, the information was wildly outdated, as you would expect.

You should always double-check any data you pull from ChatGPT and ask for sources. More often than not, the tool struggles to give a definitive source for its data, adding to the frustration. No source = unusable data.

End consumer design

Both Midjourney and Dall-E are popular AI tools for creating images from scratch. While they both vary greatly in quality, the output is a thousand times better than if you forced a PR professional to whip something up in Illustrator.

Design can be one of the most expensive and time-consuming investments for a PR campaign. There's skill in every step, whether sourcing the right designer for your vision or writing a detailed creative brief. Fall down at any hurdles, and you've made a very costly mistake. This can make it tempting to cut corners, but AI will never generate an image as well as a seasoned designer can.

As an example, AI famously struggles with hands. Here, I used the prompt “Close-up of a person rubbing their hands together. Hyper realistic. --ar 16:9”:

Not just from a quality point of view, but AI also reflects negative biases from the information it’s trained on. This means it can exhibit racist, classist, homophobic, and sexist views within its art. Any bias that a human can have, AI has too, and hyper focuses on it.

There’s also an ethical standpoint around conjuring up images of things that never happened to make a point. In May 2023, Amnesty International used AI-generated images to promote their reports on social media, and let’s just say it didn’t go down well. In general, humans are still not great at figuring out an AI image from a real one, so you have to tread carefully to be transparent when you do use AI instead of trying to deceive.

Of all the areas of PR, AI should be used the most conservatively when communicating with journalists. By nature, our jobs revolve around building strong relationships with clients and colleagues but also journalists who we hope will cover our stories.

By using AI to cut corners in this, you’ll likely harm those relationships that you’ve so carefully nurtured since the beginning of your career.

Pitching commentary from experts

One of the most common and unethical ways that AI is used in PR is the creation of fake expert comments or entirely fake experts themselves.

Expert commentary, by definition, should be coming from an expert, not an AI tool. Journalists looking for this kind of content want to work with someone who has more knowledge than them to bolster or lead a story they have on a particular topic.

Using a fake comment that no expert has had any hand in creating is not only seriously damaging to your relationship with the journalists and client but also spreads misinformation. The more serious the topic, such as health or money advice, the higher the risk is to readers.

Some PRs have even gone so far as to create phony experts with AI to push their fake expert commentary. This has included a fake bio and AI-generated image that journalists are starting to spot much quicker.

Writing press releases

Using AI to generate written content is by far its most common use. As PRs, we might find that writing press releases is one of our more time-consuming tasks, but that doesn’t mean we should immediately turn to AI.

There’s a multitude of issues that form from using AI to communicate a story to journalists. The most obvious one is that your content is likely to sound more robotic and less like it’s written by a human. While you might think it’s not obvious, it definitely is.

Digital PR also has so many nuances when it comes to writing a press release that AI just can't grasp. This includes linking to the client's website at a specific point, leading with a certain part of the story, and even considering the formatting of the release.

Many journalists are now using AI detection tools to ensure that the content they’re reading and publishing actually comes from a human. If they find out you’ve used AI, they could refuse to publish your story, or even worse, you’ve damaged your relationship and harmed the reputation of your client.

Write emails to journalists

Similar to the above, when communicating with journalists over email, you should never use AI.

Emails should be personalized as much as possible to the specific journalist, something AI can’t always do. It doesn't know that Laura has just returned to work from maternity leave or that Michael recently wrote about the rise in hidden fees of airlines. When mentioning these things, you're building strong relationships.

With the use of AI, there’s also a high chance that you’ll start to sound bland and devoid of personality. Anyone who has ever worked in PR knows that talking to journalists like the humans that they are will get you a long way.

While this overly familiar email might be acceptable when chatting colleague to colleague, time-strapped journalists would balk at the sight of it.

Find journalists to reach out to

Finding the right journalists can be a long and arduous process, but it’s one of those tasks that, the more you cut corners, the less relevant the journalist is likely to be for your story.

By asking ChatGPT, you can absolutely get a list of journalists who might be related to your niche or likely to pick up your story. At least, that's how it looks at first. Upon closer inspection, these journalists are often littered with false or out-of-date information.

For example, below, I asked for some journalists who cover home trends. To its (small) credit, the AI gave a disclaimer that we should check their current details as journalists move around a lot.

Even so, the first journalist on the list appears to be a freelance journalist, which isn’t the end of the world but probably not my top choice. When we do a little more research, we find that she hasn’t written for The Telegraph since April 2023. If we were to pitch an idea for that specific publication, it’s unlikely it will be considered for that publication. We also find from manual research that the stories she covers for The Telegraph tend to be case studies and interviews, so a story on home trends and statistics is unlikely to interest her.

Similar to the second journalist, after a quick Google of her name, it seems that she has never worked with The Guardian. In fact, her expertise lies in US politics, and she doesn’t seem to have ever written on home trends.

The best thing to do is use existing databases that are updated by real people like Roxhill. However, the best route you can go down is by manually finding each journalist yourself by searching for similar stories they’ve recently written. It takes a long time, but it keeps relationships safe and ensures every journalist has been sense-checked.

Now we know which aspects of the digital PR process we shouldn’t be using AI, what are the best practices we can keep in mind moving forward to ensure our use of AI doesn’t affect our work?

Be transparent

If you’ve put together a digital PR campaign, it’s always best practice to have a very clear methodology that states precisely how the project was put together. This gives journalists more information to help them ask the right questions.

AI should be looked at just like any other tool. If you’ve used it in a campaign, be clear about where and how it was used. This builds trust with journalists and your audience instead of trying to hide how it has benefited your campaign.

Use it to improve, not replace

One of the biggest criticisms of AI is that it will eventually replace workers in specific jobs or tasks. It should be used to help nudge you in the right direction, speed up processes, or clarify something. AI shouldn’t be used to replace anyone or any full task.

If you get to a position where you ask yourself, “Should I pay for someone to do this for me, or shall I get a watered-down version from AI?,” hopefully, you should already know the answer.

Don’t let it affect quality

AI should never, ever, ever get in the way of the quality of your work. Everything you ask an AI to do should be checked or rewritten to ensure your work's quality is as high, if not higher than before you started using it. Don’t cut corners.

Check for plagiarism and copyright issues

As AI is a trained model that takes existing information, content, and images to create something new (much like the human mind), it’s important to understand where these influences come from. More often than not, AI will plagiarize content, and you’d better double-check the information it gives you before you have a lawsuit on your hands.

The importance of a human touch

In the field of PR, the human touch remains essential and irreplaceable, especially as AI becomes more prevalent. AI should be seen as a valuable tool to complement human expertise rather than a replacement. Humans bring unique qualities to PR that AI simply can't replicate, and that’s where our power is.

One area where humans excel is in creativity and emotional intelligence. Crafting compelling narratives and connecting with audiences on an emotional level are skills that humans excel at. Relationships in PR thrive on authenticity, empathy, and adaptability, qualities inherent to humans. While AI can automate tasks, it can't establish genuine, trust-based relationships.

As long as we still value human interaction, AI will never replace the PR profession.

To achieve a balance and make sure you’re not being left behind, PR professionals should blend AI and human involvement. AI can analyze data and identify trends quickly, providing insights that humans can use to make strategic decisions. Content creation can start with AI-generated drafts, which humans can then refine with creativity and context.

By harnessing AI's strengths alongside human intuition, PR can adapt to the digital age while maintaining authenticity and emotional resonance, ensuring that the human touch remains at its core.