University of Southern Maine researchers are developing an ethics training program they hope will prevent scientists from taking shortcuts or cheating with artificial intelligence and other technologies.

Faculty and a doctoral student from USM’s Regulatory Training and Ethics Center are conducting a federally funded study to learn why people make unethical research decisions and how to train scientists to avoid them by logically processing their thoughts and feelings before they act. The study is specifically focused on helping scientists, engineers and mathematicians make ethical decisions in the era of AI and in stressful situations.

“A lot of research happens under high pressure,” said Bruce Thompson, a professor of psychology and principal at the ethics center. “A climate can start to form where people feel they need to produce results, and that can lead to plagiarism, research misconduct and data fabrication.”

Researchers are particularly vulnerable to ethical lapses using artificial intelligence – digital tools and programs that can learn and mimic the work of humans. The technology can make it tempting to take shortcuts rather than relying on one’s own work.

Thompson and other researchers at the Portland-based regulatory training and ethics center hope to better understand what is behind individuals’ tendencies to cut corners ethically and use that information to create training programs for businesses, nonprofits and colleges – including those in the UMaine System – that could help prevent cheating or other unethical conduct in research.

The ethics center received $400,000 from the National Science Foundation to develop and test the training over three years.

People have long studied questions of honesty in science and research, but looking into ethical problems in research is particularly important now with the advent of technology such as ChatGPT, Thompson said.

ChatGPT is a an AI-based program that simulates human conversation and can instantly answer questions and write text in a human-like way.

“People have always done easier things under pressure, but (artificial intelligence) amps up this problem,” Thompson said.

The project is an example of efforts happening at all levels of education and business to address the expanding ethical challenges around AI and other digital technologies. But the way the regulatory training and ethics center plans to explore the issue is unique, he said. It focuses on metacognitive reasoning – the ability to logically and mindfully reflect on the way your own brain is working.

The ethics center plans to put subjects through high-stress scenarios that could lead them to taking ethical shortcuts and test their initial ability to avoid them. Then, the ethics center researchers plan to put them through a scenario that allows subjects to reflect on their thoughts and feelings before acting. Finally, the center plans to measure whether test subjects’ tendencies to cheat have changed.

“We hope to create a level of self-awareness so that when people are on the brink of taking a shortcut they will have the ability to reflect on that,” Thompson said. “It’s a preemptive way to interrupt the tendency to cheat or plagiarize.”

The testing is to be done on USM students who are studying research design or taking research-focused classes and volunteer to participate. High-stress scenarios will include time constraints and other outside pressure.

The study and the training program that results from it will put USM “at the cutting edge of student research readiness,” Thompson said.

Students will acquire tools they can use to help them make sure their research is ethical and has integrity, the researchers said.

“It will help them manage stress that is inherent to research,” said Ross Hickey, an expert in research compliance who serves as USM’s assistant provost for research integrity and the director of the regulatory training and ethics center.

In addition to being used for college researchers, Hickey and Thompson said they hope their training can be used to train researchers in institutions that receive federal funding. Recipients of federal research funds are required to be trained in how to conduct responsible research and teach regulations, policies, conduct, standards and ethics.

Preserving ethical and accurate research is crucial to society, Hickey said.

“You want the public to have faith in research produced by scientists and researchers,” he said. “That means it has to be conducted with integrity and honesty.”

Unethical practices such as cherry-picking data and data fabrication can lead to inaccurate data and to bad policy, medical or other decisions.

Most researchers want to do the right thing, but stress or other external factors can cause people to act in unethical ways, Hickey said.

“We want people to have the tools to think about what they should do in stressful situations,” he said. “The more people that do that, the stronger the society.”

Copy the Story Link