Looking To AI for Decision-Making in Extreme Situations

Summary: Researchers are investigating the use of AI technology in complex decision-making situations with the goal to identify the best human attributes artificial intelligence can mimic when faced with making difficult decisions in extreme situations.

Source: UMass Lowell

Imagine you are a doctor managing the emergency room of a large hospital. You suddenly get a call reporting a mass shooting at a nearby concert. In 20 minutes, you will be responsible for triaging more than 200 patients with a range of injuries. You do not have enough staff or resources and the hospital policies are not designed for a situation this dire.

“When people respond to emergencies, many decisions they face are quite predictable. They’re trained on them, and there’s policy,” said UMass Lowell’s Neil Shortland, associate professor in the School of Criminology and Justice Studies.

“But every now and then, they get stuck with a really tough decision they’ve never trained for or never experienced, and don’t have any guidance as to the right thing to do.”

He added: “Although these decisions are rare, they occur in the most extreme situations with the highest stakes.”

Shortland is leading a group of UMass Lowell researchers studying the use of artificial intelligence to help make difficult decisions like the one proffered above. The team consists of computer science Assistant Professor Ruizhe Ma, electrical and computer engineering Assistant Professor Paul Robinette, philosophy department chair and Associate Professor Nicholas Evans and Holly Yanco, professor and chair of the Miner School of Computer & Information Sciences.

The U.S. Department of Defense is funding the project through a $3 million grant from its Defense Advanced Research Projects Agency, with $1.2 million going to UMass Lowell and $1.8 million to industry partner Soar Technology, Inc. The Michigan-based firm builds intelligent systems for defense, government and commercial applications.

Modeling human behavior

The goal of the research is to find the best human attributes that AI can mirror when making difficult decisions in extreme environments.

“We’re harnessing the essence of a person by modeling them as their best self,” Shortland said.

Human judgment is fallible. Even if someone is highly qualified to make a decision, their judgment can be skewed by biases, hunger, tiredness and stress among other factors, he explained.

“AI eliminates those issues. It can be the best version of a person each time,” Shortland said.

AI also helps increase the number of decision makers in situations like mass shootings, where instead of having just one doctor assessing victims, dozens of robots could be deployed to evaluate the victims after being programmed with AI that models the doctor’s decision-making processes.

This shows the outline of a computerized head
The goal of the research is to find the best human attributes that AI can mirror when making difficult decisions in extreme environments. Image is in the public domain

To study the best human attributes for different decision-making scenarios, the researchers will expose people to emergency situations using a computer research tool developed by Shortland called the Least-Worst Uncertain Choice Inventory For Emergency Responses (LUCIFER). They will then measure how a person’s psychological traits and values impact their decisions.

“When we identify the key decision-maker attributes, we will be able, to some extent, quantify a decision process and develop AI decision systems tailored to specific needs and environments,” Ma said.

One scenario the research team is focusing on is triaging patients. Using LUCIFER, test subjects will be presented with visuals of patients with various injuries and vital signs before determining if they are OK, if they are eventually going to need medical assistance, if they need help right away or if they are deceased. 

The researchers are also developing a 3D simulation that immerses test subjects in triage scenarios.

“The triage micro-world will allow us to evaluate the progress of the overall project,” said Robinette, who is designing the 3D simulation with his students. 

“It will help us see if what we’re finding in our LUCIFER studies transition into a more real-world environment,” Shortland added.

For the research project, the team will be utilizing on-campus resources, including UMass Lowell’s Misinformation Influence Neuroscience and Decision-making Lab and the New England Robotics Validation and Experimentation Center, while tapping the researchers’ range of skills and expertise.

“Interdisciplinary teams are required to push research out of the lab to the real world, where it can save lives,” Robinette said. “I’m looking forward to the great things we can all do together.”   

About this AI research news

Author: Nancy Cicco
Source: UMass Lowell
Contact: Nancy Cicco – UMass Lowell
Image: The image is in the public domain

Join our Newsletter
I agree to have my personal information transferred to AWeber for Neuroscience Newsletter ( more information )
Sign up to receive our recent neuroscience headlines and summaries sent to your email once a day, totally free.
We hate spam and only use your email to contact you about newsletters. You can cancel your subscription any time.