Can Artificial Intelligence help prepare for the next crisis?

ChatGPT is a state-of-the-art language model developed by OpenAI, a leading artificial intelligence research lab. The model is based on the transformer architecture and is trained on a large text set, allowing it to respond to natural language prompts with highly coherent and fluent answers.

One of the key features of ChatGPT is its ability to understand and respond to context, allowing it to carry on a conversation in a way that feels natural to users. This makes it a valuable tool for a wide range of applications, from customer service chatbots to content creation.

ChatGPT's potential for use in the field of communications is significant. It can be used to generate written content such as articles, blog posts, and social media posts, freeing up time and resources for communication professionals. Additionally, it can assist in the creation of scripts and dialogue for videos, and even help with language translation tasks.

ChatGPT, like any machine learning model, has certain limitations.

Some of these limitations include:

1. Lack of common sense: ChatGPT is trained on a large dataset of text, but it does not have a general understanding of the world like humans do. This can lead to it providing nonsensical or irrelevant responses to certain prompts.

2. Bias in the training data: As ChatGPT is trained on a dataset of text, it can inherit the biases present in that data. This can lead to it generating responses that are insensitive or discriminatory towards certain groups.

3. Limited understanding of emotions: ChatGPT can generate text that appears to convey emotions, but it doesn't have the ability to truly understand the emotions behind the text.

4. Difficulty with open-ended questions: The model is trained to generate a specific response to a given prompt. If a question is too open-ended or doesn't provide enough context, the model may have difficulty generating a coherent response.

5. Limitations in the quality of the answer when the question is not well formulated, the context is not clear or the knowledge cut-off date is surpassed.

6. The model is still not perfect and can be influenced by the quality of the input it receives.

It's important to keep these limitations in mind when using ChatGPT and to always verify the output it generates before using it in any real-world applications.

Use cases for ChatGPT in the context of Crisis Communications

After using ChatGPT for several weeks, it is clear that it can be used to support crisis communication preparation in the following ways:

Supporting the development of crisis simulation scenarios.

"It is a dark and stormy evening in Amsterdam North. The wind blows hard and the rain beats against the windows of the houses. Suddenly the population hears a thumping sound from the IJsselmeer. A dike breach has occurred and the water is rushing in."

This is the intro of a realistic crisis scenario that Tom Compaijen created with a single prompt in ChatGPT and then used the rest of the scenario for training.

Drafting correct prompts is crucial, and this is where human experience and knowledge of the subject - in this case crisis communication - make all the difference.

I asked ChatGPT to first draft a possible scenario based on a cyber-attack on a fictive company called SureIT, and then write a first reactive statement. You can see the results and the prompts used here.

Drafting reactive statements.

When a crisis hits you do not have time to stare at a blank page. Here ChatGPT can play a role in drafting a first version of a reactive statement based on the facts and hints you give it. Once you have the result, you should of course correct and adapt it according to your corporate communication policy.

My prompt was: please write a first reactive statement the management of SureIT will distribute to the media. The tone needs to be reassuring, transparent and supportive of their customers. (Note the tone hints)

The result was the following:

"We are deeply sorry to inform our valued customers that SureIT has been the victim of a cyberattack resulting in the theft of sensitive customer data. We understand the seriousness of this incident and want to assure our customers that we are taking immediate action to address the situation.

Our IT team has been able to contain the attack and is currently working to determine the extent of the data breach. We are taking all necessary steps to protect our customers' personal information and prevent such incidents in the future.

We want to inform our customers that we will be reaching out to them directly to provide them with information on how to protect themselves from potential fraud. We will also be offering free credit monitoring services to affected customers.

We want to assure our customers that the safety and security of their personal information is of the utmost importance to us. We are committed to being transparent and informative throughout this process and will keep our customers informed of any developments.

We are also cooperating with law enforcement and forensic experts to determine the cause of the attack and identify the individuals responsible.

We apologize for any inconvenience this incident may have caused and want to assure our customers that we are taking all necessary steps to protect their personal information. If you have any concerns or questions, please don't hesitate to reach out to our customer support team.

We thank you for your patience and understanding during this difficult time."

Drafting key messages.

Again, in the same context and with the same background information I gave ChatGPT about the cyberattack on SureIT, I now asked it to create 3 key messages which could be used with the media and other stakeholders.

Here are the results:

• "The safety and security of our customers' personal information is of the utmost importance to us and we are committed to being transparent and informative throughout this process."

• "We are cooperating with law enforcement and forensic experts to determine the cause of the attack and identify the individuals responsible, and taking steps to prevent similar incidents in the future."

• "We are taking immediate action to address the situation and protect our customers' personal information."

Outlining a step-by-step approach for a crisis communication plan.

Next I asked ChatGPT to write an outline of a potential crisis communication plan based on the facts provided. You can see the results here.

Interesting to note is that ChatGPT also takes employees into account. At least the AI and I are on the same line; first communicate internally, only then externally. Employee communication is often forgotten in crisis communication plans. Well done ChatGPT!

I also appreciate it that ChatGPT stresses the need to “Communicate promptly and transparently” and “Evaluate and improve”. Again, many organizations do not use the end of a crisis to review policies, SOPs and crisis communication tactics. A crisis should be used as an opportunity for learning and improvement, and to make necessary changes to prevent similar crises in the future. Another good point made by AI.

Listing best practices when you need them.

ChatGPT can also quickly give you the top best practices of certain crisis communication approaches based on the data it has to analyze.

I asked: “List the best practices based on the situational crisis communication theory by Timothy Coombs”.

Here’s what it gave me in return:

• Anticipation: Prepare for potential crises by identifying and assessing potential risks, and developing a crisis communication plan.

• Speed: Respond quickly to a crisis, as delays can escalate the situation and damage an organization's reputation.

• Transparency: Be open and honest in communication, and provide accurate and timely information to stakeholders.

• Credibility: Establish and maintain credibility with stakeholders through consistent and reliable communication.

• Empathy: Show empathy and understanding towards those affected by the crisis, and address their needs and concerns.

• Control: Take control of the situation by managing the crisis and its communication, rather than allowing the crisis to control the organization.

• Constancy: Be consistent in messages, tone and actions throughout the crisis and its aftermath.

• Coordination: Coordinate crisis communication efforts with relevant internal and external stakeholders, such as government agencies and the media.

• Evaluation: Continuously evaluate and adjust the crisis communication plan and response as needed.

• Learning: Use the crisis as an opportunity for learning and improvement, and make necessary changes to prevent similar crises in the future.

I’ll be hosting an IARCC webinar with Timothy Coombs soon and will ask him to review the answers… 😅

Media relations “Red Teaming”.

In the parlance of the cybersecurity sector and the military, a “red team” plays the role of an adversarial entity, to probe an organization’s defenses and find susceptibilities to breaches.

In this case, I asked ChatGPT to act as a reporter from the Wall Street Journal and list the top 5 questions he would ask the PR Manager of SureIT based on the given facts.

This approach can help with brainstorming the media briefing and preparation of spokespeople in the context of crisis communication.

Here are the questions ChatGPT - now playing the journalist - gave me:

1. "Can you provide more details about the nature of the cyberattack and the extent of the data breach? How many customers have been affected and what kind of personal information has been stolen?"

2. "What measures has the company taken to contain the attack and prevent similar incidents in the future? Are you confident that the stolen data is secured?"

3. "How is the company supporting and communicating with affected customers? What kind of assistance, such as credit monitoring services, is being offered?"

4. "What is the company doing to cooperate with law enforcement and forensic experts to identify those responsible for the attack and bring them to justice?"

5. "Can you comment on the potential financial impact of this incident on the company and its shareholders? Has the company taken any steps to mitigate this impact?"

I like this approach very much as it gives a crisis communicator a good idea of which questions to expect (as long as the prompts are well crafted).

What’s next for AI-supported Crisis Communications?

Although it caused quite a stir when it became publicly available, ChatGPT is just one AI tool that can be used by communicators. To write this article, for example, I used ChatGPT, but also Instatext (English is not my first language) and ChatSonic.

Other tools based on AI and machine learning can already help us create text, audio and video content in the shortest possible time (think of the role time plays in a crisis) or perform predictive analytics before a crisis occurs.

Will AI tools replace professional crisis communicators? No, but it will make those who do not understand the role of AI redundant. As someone wrote recently: “you will not be replaced by AI, but by other professionals who know how to use AI”.

I look forward to reading your comments!

If you liked this article, why not subscribe? Plus, I’ll invite you to the first virtual mini conference on the topic of AI and Crisis Communications we’re organizing with a couple of professionals. It will be free and fun to discuss the future.

You can also “collect this article for a small contribution. But that’s really up to you.