ChatGPT is now back up and running after an outage floored the service for a large part of the working day (4 June). 

The headlines spoke volumes about the chaos it caused. “ChatGPT outage leaves thousands unable to use the chatbot,” screamed the Metro, while Gizmodo went with “ChatGPT back up after going on the fritz.”

But the headline that totally skewered the story was delivered by The Register with its usual sass: “Millions forced to use brain as OpenAI’s ChatGPT takes morning off.”

People took to X —  formerly Twitter — to share their dismay with one even asking: “How am I supposed to get any work done with ChatGPT being down!?”

Outage caused panic 

Make no mistake — ChatGPT and other generative AI platforms are having an enormous impact on the way people approach creative work that involves words, images and sounds. 

But you only get a real sense of the impact it’s making when things go wrong — or in this case, it falls over altogether. 

Everyone knows AI tools have the potential to revolutionise the way we work. And the world of PR and comms is one of those industries where it’s certainly making its mark. 

Here at Wildfire, we’re having conversations about AI with clients on a daily basis. That’s why we established an AI taskforce to help make sense of all the noise. 

Let’s be clear. ChatGPT and other tools are great for a quick and dirty bit of research. Or a swift explanation of a difficult-to-grasp topic. It’s handy for a spur-of-the-moment brainstorm if no one else is around. And it can bash out a rudimentary story skeleton in next to no time.

As a writing tool, it’s a neat trick — a four or five out of 10. Maybe a six on a good day. But that’s it. And as The Register’s headline makes perfectly clear, it is no substitute for good old brain power, creative thinking and editorial know-how.

Despite all the rhetoric when it comes to writing copy that makes an impact, such AI writing tools are lacking. For now. But there is one area where they do excel. They’re a really good editorial assistant. Possibly an eight or nine out of 10 depending on the task. For instance, AI is great for: 

• Generating headlines

Simply slap in some text and ask it to come up with a headline. Easy. Even better, ask it to come up with a dozen headlines so you have a choice. Even if it doesn’t come up with a winner the first time around, sometimes it’s enough for you to cherry-pick from the ideas generated to come up with your own. In a sense, it’s like bouncing ideas off a colleague. 

• In short, it’s good at précis 

Being able to précis longer text has always been a valuable — if somewhat overlooked — skill. Journos are familiar with having to précis reports, speeches, and documents to provide readers with concise and clear news articles. Likewise, in business, execs often expect a snappy version of a report or article to make their life easier. Conversational AI is great at this. Slap in the text, set a word count, and press ‘go’. Job done. Just remember to check the output to make sure it hasn’t missed anything important. 

• Metadata

Similarly, if you’re writing blogs and need to provide the metadata as well, it’s well worth thinking about using ChatGPT or Google Gemini. They should be able to pick out some SEO keywords as well. 

In each of these cases, ChatGPT and other conversational AIs come into their own as an editorial assistant. They take some of the drudgery and legwork out of editorial work. But it would be unwise to carry out any of these activities and publish what these tools have created without checking first. In other words, writers using these tools as editorial assistants must then become sub-editors and editors to ensure content passes muster.

 Another interesting area where the technology comes into its own is research. 

• Explaining new ideas, concepts and technologies. Bringing you up to speed about recent events.

If you need to know what quantum computing is, for example, just ask. And within seconds it will start reeling off something like: 

Quantum computing is a type of computing that harnesses the principles of quantum mechanics to process information. Unlike classical computers, which use bits to represent data as 0s or 1s, quantum computers use quantum bits, or qubits, which can exist in multiple states (0, 1, or both simultaneously) due to superposition.

If such answers leave you bamboozled, you could always try this neat hack and add ‘ELI5’ — Explain Like I’m 5’. 

Quantum computing is like having magic coins that can be heads, tails, or both at the same time. These coins can also be magically connected so that flipping one instantly tells you the result of the other, no matter how far apart they are. This lets quantum computers solve super hard problems really fast.

To be fair, even the answer for five-year-olds is still a head-scratcher. But at least it’s a start. 

Chatbot etiquette

Be in no doubt — generative AI is an amazing technology. The more we use it, the more we’ll begin to understand its strengths and weaknesses and how to get the best from the technology. 

We’ll also begin to be able to refine the prompts to improve the output. In fact, it will be like training a new member of your team. 

But if that’s the case, should we be polite when we ask questions and give it tasks? Would AI do a better job if we said ‘please’ and ‘thank you?’ 

“You can be as polite as you feel comfortable with when interacting with ChatGPT,” chirped the chatbot when we popped the question. 

“Including polite phrases like ‘please’ and ‘thank you’ doesn’t necessarily improve the responses you receive in terms of accuracy or relevance, but it’s always nice to observe common social niceties, even in digital interactions,” it said. 

Quite right. Noted.