“The speed at which technology is developing makes it difficult to make credible predictions.”Now take that statement, change your perspective on it and write: “given the speed of technological innovation, predictions need to be recalibrated every 8-12 months”. These statements say the same thing but hit different: the first is defeatist and expresses uncertainty and lack of clarity, the second is authoritative and clear-headed. They could have been generated by AI by the same prompt. The difference lies in the end strategy, the definition of what you want to convey with this statement. It’s about having an eye on an end game.
This end game is typically positioning, which has been formulated beforehand, at board level, with the support of your communications team and PR agency. Unlike AI, your comms team and PR agency understand that positioning your company as reliable and confident despite market volatility is an ongoing task and that messaging needs to be reflected in every communication. Unless you remember to include that in every prompt, your communication won’t hit the mark. Unlike an LLM, your agency doesn’t need reminding to reflect a blend of nuanced messages at every turn but instead infuse everything they write for you with intentionality.
Overconfidence in the power of LLMs is understandable, and it is also true that more experienced users are able to extract more fine-tuned results, but the reality is that in most businesses, users are putting a request into a model without necessarily providing all the information that is required to receive an accurate output. Anecdotally, a client once asked AI to draft a release that needed some templated elements in order to secure speedy approvals. The LLM reworded those because it underestimated the need for adherence. The humans involved caught this just in time.
I’ve been playing around with AI for a while and I’m delighted with the time saving and efficiency it offers to communications, even without a subscription. For example, I can ask it to provide an initial batch of research, I may ask it to give me concrete examples of how a specific technology is used within a sector I don’t know well, and I will most definitely ask it to proofread this article for grammar once I’m finished. What I would never ask it to do is write my content for me, pitch to journalists or get involved in my professional relationships.
The fact is that PR is about building three key relationships: with audiences, through savvy, well balanced content; with clients, by getting to know the personal dynamics and long-term objectives of the business; and with journalists, by talking about what they would like to receive and understanding how we can help them.
When it comes to the relationship with the client, a PR professional worth their salt will know (sometimes even sense!) where the bodies are buried and steer well clear. This is not the type of information you put in writing so it’s hardly surprising no one would feed it to an LLM. Client relationships go beyond listing pet-peeves and preferred terms. Your PR agency should understandinternal politics, who not to quote in a press release and who to work with instead, perhaps because they are faster in approvals, because they have a greater depth of expertise, or because they work with a specific sector or market. This type of understanding is what builds trust between a PR agency and their client. It’s about having your back even when it is against their interest. You can’t codify trust and so you can’t build it with an automation.
Another AI press release debacle I recently witnessed saw the insertion of a super-promotional paragraph about the company into a press release. Any PR professional will tell you that while you might find this type of press release online (and presumably that’s what the LLM learned from), it’s terrible PR and no journalist will include it in their story. Similarly, the LLM offered a very zingy but non-strategic quote that did not tie into any particular position and took no account of the speaker’s particular tone of voice as it did not have access to the wealth of interactions and conversations the PR professional had taken part in, which revealed the spokesperson’s down-to-earth and warm style of communication.
The non-writers will argue that an endless set of data inputs and insanely long prompts could fix all these defects, but this lengthens production times so at that point, why bother? When any PR manager can do this just as fast using their brain and empathy?
Empathy is another hurdle where I see AI crashing in communications. People struggle to simulate it, so how can we expect our audience to believe it when it comes from a machine? Crisis management in particular requires empathy, modulation of messages and a human touch. Can you imagine the ethical implications and reputational impact of communicating with the victims of a toxic material leak via what is essentially a bot?
Some have argued that AI can be a useful tool in media outreach. It can certainly help rationalise the admin aspect but does it have the experience to tell you this particular journalist can be tricky to deal with? Whether an editorial team is a good fit for the client or not? It’s about matching how people vibe as much as the topic of conversation.
After outreach traditional, human PR people tend to pick up the phone and follow up with journalists. Who on earth is going to engage with an automated calling system to explain what they are working on and what they would like to see more of? So many of these conversations take unexpectedly positive turns based on the ability of the professional to think up a more suitable angle on the fly, offer another client or another piece of content.
Finally, I’ve tried some of the leading coverage tracking systems that claim to use AI and have been disappointed. Systems fail to pick up physical items of print coverage, are unable to rate a rapid mention on wire compared with a top ten listing in a decent trade publication and have more than a few difficulties when multiple languages are involved. They are also unable to consistently tell the difference between a piece of paid advertorial and earned media. When a human places a story they follow it up and are sure it comes out. Sometimes the relationship is so good the journalist emails the published piece reinforcing the positive relationship.
I wholeheartedly welcome AI into the life of the PR agency but its role is to make things more efficient and faster, not to replace the experience, breadth of knowledge and interpersonal nous of professionals. It’s a great intern for now. Check back in 12 months please.
