There was a question this week that made me think. Broadly it’s this: “What do I do when people senior people to me come to me and think ChatGPT can do all my comms now.”
ChatGPT is easy. So is Google’s Bard. Ask a question and hit the button. Seconds later you’ve got a plan. What’s not to like?
For me, ChatGPT, is a spare brain. It can fill, a blank page in seconds and some of those ideas will be really good. Maybe 20 per cent will not but it takes a human to spot that 20 per cent.
If you’re the Leader of a Council with not much experience of comms a comms plan knocked up by ChatGPT is going to look pretty plausible.
So what do you do?
The question here is absolutely fundamental to how AI will reshape jobs.
If you’re buying a house you’d be advised to commission searches to see what planning applications, rights of way and mining your dream plot is affected by. The key information is largely stored on paper files that searches clerks must go through by hand. There will be few who will mourn the loss of these clerks but if its you will.
What gives comms a God given right to communicate?
There is no God given right. You need to demonstrate that your experience and advice is sound and that you’ll reach an audience, that they’ll do that thing and the organisation will be helped.
Ban it?
Of course, the first temptation is to ban tools like ChatGPT. Some UK Universities have indeed outlawed the use of large language models. They can do this because they have the power to grant or deny a degree. But what worries me is that this won’t make it go away.
Banning technology? Until 2011 no police officer in the West Midlands could use the internet because a previous Chief Constable hated it. Hindsight shows us this is ridiculous. So the real question is, not that it should be banned but how can it be used appropriately, safely and effectively?
The ClipArt argument
The first clipart was introduced by IBM in 1983. The VCN ExecuVision package allowed users to add arrows, boxes and other basic images to their work. It was a revolution. The upside was that basic design was freed from the tyranny of graphic designers. The downside? it often looked sh-t. Then people came back to designers for the really important stuff.
The effectiveness of clipart was tested out in real time and at fairly low risk.
The argument of allowing ChatGPT to produce comms so the budget can be cut by a third is outwardly compelling but what would be lost by minimising human intervention?
Political nouse
Nouse is an untaught filter to evaluate actions based on experience and knowledge. AI doesn’t usually have it and humans do.
For example, don’t turn up as a local government comms person on election night with a ‘Vote Labour’ sticker on your jacket, Why? Because you’re supposed to be politically neutral and questions will be asked.
In my time in local government there was only one person I worked with in comms who had no political nouse. They’d been a journalist and genuinely didn’t see the difference between being a journalist and a press officer. The individual would laugh and joke with opposition councillors before Full Council and ignore the other members. It wasn’t long before questions were asked not just of them but of the wider team. It caused a lot of problems and as we dug into it there were many more examples of a lack of nouse by this individual.
ChatGPT may produce a comms plan but it may also suggest materials that your judgement says you wouldn’t be using for good reason. Maybe because that particular suggestion wouldn’t fly. Or maybe because it suggests a course of action that’s too expensive.
I asked ChatGPT to produce a script for Staffordshire Day. It did. It suggested shots of Stafford Castle, Alton Towers and Cannock Chase. Excellent. But the final shot of crowds each with Staffordshire flags and fireworks just wasn’t going to work without time and a big slab of budget.
There’s no way I could have made that video unedited and here’s where comms can add value.
The costed ChatGPT test
AI tools can be useful but as the CIPR’s ‘Humans Needed More Than Ever’ research shows humans are still needed. The Guardian newspaper is closely restricting the use of AI. UK Government is considering what restrictions to put in place to ethically and safely use AI. We are not at the stage of putting out foot down on the accelerator. Buzzfeed sacked most of its nerwsroom and was rewarded with a jump in share price. Can a council do the same? It can, but it’ll miss out on the skills of human operators to sift the good ideas and to use nouse to navigate away from the bad.
However, the language of pounds shillings and pence is definitely part of the landscape for AI.
So, here’s a test. Create a comms plan for a topic the traditional way. Then create a comms plan using a tool like ChatGPT. What good ideas did it have? What’s the full cost in staff time, videography, photography and other resources of that ChatGPT plan? What’s the risk to political nouse? From all this you can see the financial and creative benefit of ChatGPT, the cost of using unbridled ChatGPT and the risk of removing political nouse from the equation.
“The comms plan we made will take two days of staff time to deliver. We’ve managed the political risk of spending too lavishly on this.
“The ChatGPT comms plan will cost £3,000 and five days of staff time. The political risk is FOI on what the cost is.”
“However, some of the ideas ChatGPT came up with we’ll repurpose.”
In other words, you’re testing it, you’re assessing the benefits, you’re calculating the cost and the political risk and THAT is the conversation to have with senior people.