BAPCO 2024 speaker focus: Keri K Stephens

The University of Texas professor in organisational communication and technology discusses emergency services’ attitudes to – and hopes for - AI.

Keri Stephens

What’s been your experience of working with emergency services organisations up until now?

I have teamed with several emergency managers to better understand how we can use social media, generative AI, machine learning, and model data to help them be better prepared during emergencies. I have written a study, which is a good place to learn about some of the work my teams have done. 

I also team with emergency managers to develop better ways to communicate about disasters that will resonate with their communities and stakeholders. For instance, my team collected data with emergency responders on wildfire preparedness.

I have received National Science Foundation funding - with an emergency manager as a CIVIC partner - to be sure the research we do is grounded in the types of knowledge that emergency services organisations need. 

What’s your sense of how first responders view AI and its potential?

First responders vary in their responses to AI. Some of them are really digging in and trying to figure out how to use it to improve their efforts.

The field as a whole can be fairly cautious, for good reason. They hold lives in their hands, and want to be sure they can use technologies that will work when most needed.

How is the technology currently being used, and how is that use broken down across different public safety organisations?

There is a huge range of how AI and other technologies are being used. For instance, mining social media data to better understand public sentiment and identify or confirm problem areas. AI and machine learning work in the background in relation to this and these efforts are still fairly labor intensive.

Another example is having emergency responders team with AI systems, so that human expertise is leveraged to train AI systems and help them identify things that matter.

Also, using tools like ChatGPT to [come up with ideas], assist in writing grants and reports, and generate graphics for reports and outreach efforts. Different activities [in this realm] are emerging daily.

There are also teams using robots for rescue and monitoring, but that is outside the scope of the work I do.

What reservations do first responders have about the technology?

They want to use technologies that have been proven to work when they need them. They are cautiously optimistic in many situations.

What’s your view on the role of the human being in decision making? Can you see a time when the ‘human element’ is removed altogether from certain operational areas/tasks?

I see humans and human expertise as being essential for decision making around public safety. We call this process humans-in-the-loop, and many technologies are not able to make decisions around human safety to the level that human experts are. 

It is important for people to realise that in daily tasks, AI systems are working in the background already to assist with decision making. For well-defined tasks and activities like performing an internet search, AI systems can sometimes work alone if they are pulling from the right types of data and are trained very carefully.

But realise that many people are involved in keeping those systems updated and constantly evolving. Today, AI-type technologies are being found to make mistakes. For instance, cameras can be sensitive to light levels, and many of the sources of data used to train AI systems contain incomplete or biased data.

Many people are working on these issues, and I’ve already seen improvement. But what is essential right now is to teach people that these systems make mistakes, and they cannot blindly follow the output. Human decision-making processes are different from AI-decision making. 

What do you hope your audience in Coventry will take from the session?

While I will talk some about AI technologies in emergency communication, I will also share examples of other technologies being used like online apps and text message alerts.

I’ll discuss how important it is that we understand our public and communicate in ways they can comprehend, and in ways that build trust. I’m excited to share some research and practical take-aways so everyone can leave the session more informed and more cautiously optimistic about our future.

The BAPCO 2024 conference programme is available here.