The is a lot of buzz about ChatGPT.
The artificial intelligence chatbot - trained to interact in a conversational way - is already being put to the test in everything from writing articles and songs to language translation.
In one case, a recruitment team unwittingly recommended the language model for a job interview after it completed an application task.
And the publishers of the Daily Mirror are exploring whether it could write routine traffic and weather stories in its local titles.
But what about crisis communication? Could ChatGPT - the fastest-growing web app in history - and AI have a role to play here?
Don't have time to read this blog? You can listen to it here:
Well, let’s look at one of the first examples of where it has been used in crisis media management.
Officials at Vanderbilt University’s Peabody College office of equity, diversity and inclusion sent a message following a Michigan shooting that killed three students.
It said the incident was a reminder of the importance of creating an inclusive environment.
It said: “One of the key ways to promote a culture of care on our campus is through building strong relationships with one another.
“This involves actively engaging with people from different backgrounds and perspectives, listening to their stories, and showing empathy and support.
“We can look out for one another by noticing signs of distress and offering support to those who may be struggling with mental health issues.”
It went on to discuss promoting a “culture of respect and understanding” and creating a “safe and inclusive environment.”
And then there was a line - in much smaller text - at the end of the message.
It said: “Paraphrase from OpenAI’s ChatGPT AI language model, personal communication.”
And that sentence has grabbed attention:
After Michigan shooting, one university used ChatGPT to offer help to students. It backfired Los Angeles Times
Vanderbilt apologizes for using ChatGPT to write message on MSU shooting Washington Post
Vanderbilt University apologizes for using ChatGPT to write mass-shooting email CNN
Vanderbilt University officials suspended for using ChatGPT to send memo to students impacted by mass shooting Independent
Damning stuff.
To be fair to the university, it responded quickly to the backlash.
Camilla Benbow, Peabody College’s dean, released a statement saying the university would investigate the decision to write and send the message.
She added: “I am also deeply troubled that a communication from my administration so missed the crucial need for personal connection and empathy during a time of tragedy.”
The “deeply troubled” part ironically sounds robotic. But we also learnt two people have been suspended while investigations are carried out.
Using AI to write about human tragedy does not sit comfortably. It is not a good look. And while we don’t know how much was written by AI, it creates the impression leaders were too busy to use their own words.
But would people have noticed if the university had not been so transparent about using ChatGPT?
Perhaps not.
But then again, the language was vague and not personable.
During a crisis, organisations must show a human touch and personal connections.
Words should be meaningful.
There has to be a purpose beyond saying the words you think people want to hear.
On our crisis communication training courses, we use the acronym CARE - Compassion, Action, Reassurance and Examples.
And there is a reason compassion comes first.
As Theodore Roosevelt famously said: “No one cares how much you know until they know how much you care.”
And that compassionate communication is what is missing in this ChatGPT offering.
One argument I’ve seen in favour of its use in crisis communication is that it can be used to create the first draft of crisis statements. After all, when a crisis strikes, you don’t have time to stare at a blank page or suffer writer’s block.
Could it provide some support?
But these statements - holding statements - should be written in advance as part of your crisis communication preparation.
And if you are reading this now and thinking your organisation doesn’t have a bank of pre-prepared draft crisis holding statements ready, I suggest that you drop all the proactive PR you are working on and get this sorted.
And those holding statements should be based on CARE.
Even in the early stages of a crisis, organisations must show concern and sympathy for those affected, whether it is people who have been physically injured or customers unable to access their accounts because of a computer glitch.
Putting it to the test
But I wanted to test the technology for myself.
So, I asked ChatGPT to write a statement about a fictional workplace accident where one person had died and another was in hospital.
Here’s what it said:
“We are deeply saddened to inform you that there has been an accident at work resulting in the loss of one of our team members. Our thoughts and sympathies are with their family and loved ones during this difficult time.
“We are also aware that another team member has been hospitalized as a result of the accident. Our priority is their wellbeing and we are working closely with medical professionals to provide them with the care and support they need.
“We are currently investigating the incident to understand what happened and to identify any measures that can be taken to prevent such accidents in the future. We are cooperating with relevant authorities to ensure a thorough investigation is conducted.
“We understand that this is a difficult time for everyone involved, and we are committed to providing support to our team members and their families as they deal with the impact of this tragic event.”
What do you think?
I’ll be honest. It’s better than I thought. And parts of it are ok. But it feels dry and stilted. And overall, you would probably put it in the ‘iffy’ category.
“Deeply saddened” is a horribly generic start. Later, an incident where someone has died and another is in hospital is described as a “difficult time”. And the support being offered comes too low down for my liking.
Emotion is conveyed. But it doesn’t read as if that emotion is understood. Not surprisingly, it lacks the human touch.
Ethics
Speaking of the human element, we should not lose sight of the fact ChatGPT has reputational issues of its own, with TIME magazine reporting it was built with contributions from Kenyan labourers paid less than $2 per hour.
And that these workers were exposed to graphic content to clean the platform of violence and hate speech.
A spokesperson for Open AI, which owns ChatGPT, said: “We take the mental health of our employees and those of our contractors very seriously.”
Which sounds like something AI would write for you.
Elephant
But there is also an elephant in the room we need to address.
Doesn’t it make sense for someone who writes for a living to rubbish AI writing? Why would I say much in support of a machine that could take over?
Well, I think there is a lot to be said for tools that make writing easier and often use the likes of Grammarly and the Hemmingway Editor.
We mention many others during our writing skills training courses. And perhaps, in time, ChatGPT will be added to that list.
But when it comes to empathy during a crisis and creativity (show me a robot that could write about Prince Harry’s unusual willy remedy and link it to communication training), humans still rule.
Media First are media and communications training specialists with more than 35 years of experience. We have a team of trainers, each with decades of experience working as journalists, presenters, communications coaches and media trainers.
Click here to find out more about our crisis communication training and writing skills courses.
