Writing is at the heart of everything we do as a public relations agency. It’s no stretch to say we write thousands of words each day, crafting everything from story pitches to press releases to op-ed pieces and everything in between. That doesn’t even count emails, Slack messages, status updates and other written notes that are created as a matter of doing business.

Like most industries, we’ve watched the development of generative AI, particularly ChatGPT, with great interest. It was a bit frightening at first — with the right prompt, AI can turn out a pretty good draft of a blog post or news announcement.

But pretty good isn’t good enough for us, and it isn’t good enough for our client partners. Good writing has never been more important, and the ability to rise above AI-generated noise is one of our differentiators. Still, we know a good tool when we see it, and generative AI is a powerful ally in jump-starting the creative process. Here are a few ways we’ve used ChatGPT:

  • Creating outlines for bylines: “What are three ways companies can balance digital transformation and sustainability efforts?”
  • As a research toolkit and idea generator: “What are some ways that IoT and edge computing impact the automotive industry?”
  • Creating first drafts of pitches and press releases.
  • Creating headline ideas: “What are five clear, clever headlines for this topic?”
  • Checking our work: Run an approved draft through ChatGPT and ask it to identify the key message.

Creating Guidelines for Working With Generative AI

The results have been interesting, and we’ve learned a lot from using generative AI for everyday tasks. But it was also clear early on that the agency and our clients would benefit from a set of rules to guide our work. We looked to our agency values for inspiration, including transparency and authenticity. We came away with this framework to maximize productivity while minimizing risk.

  • Accuracy: We fact-check everything. ChatGPT doesn’t cite sources, so any statements it produces must be checked. Beyond the rare instances of ChatGPT “hallucinating,” it’s not uncommon for information to be plagiarized from an unreliable source or even from a competitor. Don’t take the output at face value.
  • Privacy: We never use undisclosed client information. Everything put into a ChatGPT request helps train that AI model, so we never put anything in there that isn’t public information. Our clients trust us with some of their most confidential information about their business, people and products, and we always honor that trust by keeping that information out of ChatGPT.
  • Humanity: Any ChatGPT result can be improved. We never copy-and-paste the output of generative AI and share it directly. ChatGPT does not understand context — and only we know the details of our client’s brand voice, audience and objectives. That’s why we always add meaning and nuance.

Economist and professor Richard Baldwin told the 2023 World Economic Forum's Growth Summit, "AI won't take your job. It's somebody using AI that will take your job." That’s a healthy way to look at it. Ignoring generative AI really isn’t an option, so learning how to make the most of these tools while operating within sensible guidelines makes a lot more sense.

We’re excited to expand our use of generative AI and watch as our AI clients continue to innovate in the hottest space in tech right now.