Sometimes the right opportunity comes along and changes your trajectory, opening doors you otherwise might not have envisioned. Alex Barone, CLA ‘24, has seen this play out firsthand.
As an undergrad psychology major and volunteer in Temple’s Control and Behavior Laboratory (CAB Lab), Barone participated in a research study examining what makes humans able or unable to identify AI-generated web content. The results were accepted for publication in Scientific Reports. This project, which helped set Barone on his path to grad school, was made possible by a Liberal Arts Undergraduate Research Award—otherwise known as a LAURA.
The LAURAs grant a $2,000 stipend to a student-mentor duo for a semester while they conduct a research project. The award affords students the opportunity to gain valuable research experience while alleviating some of their financial burden. For Barone, the idea to apply sprung from a fortuitously timed email.
“I actually didn’t know about the LAURA program until I received an email for the deadline, which is what pushed me to ask about making the project into something that would qualify for a LAURA,” explains Barone.
Jason Chein, Principal Investigator of CAB Lab and the paper’s lead author, decided he wanted to explore generative AI after noticing the growing conversation around ChatGPT and text generators. “I pitched the idea for this study to the entire lab group—about 20 people,” recalls Chein. “Alex jumped up and said, ‘I’ll work on that.’”
“When Jason mentioned the idea in lab meeting, ChatGPT was just starting to really take the world by storm,” says Barone. “I’d been following generative AI for a little while at that point, and when the opportunity arose to utilize it within an independent study, I became really interested in the idea.”
The team, rounded out by PhD student Steven Martinez, quickly began developing materials. The study gathered responses from 194 online participants who were asked to identify social media posts as human-written or AI-generated. Additionally, participants were put through a short set of tasks and questionnaires to broadly gauge factors like executive functioning ability, nonverbal fluid intelligence and social media habits.
One result that the team found particularly interesting was that high fluid intelligence scores were the strongest indicator of ability to discern AI from human text. On the other hand, those who spent the most time scrolling on social media struggled. Barone hopes to see future research build on these findings, and he is grateful to have had such a hands-on role in developing and executing the study.
“Alex was an invested volunteer undergraduate working in my lab before the LAURA, but he hadn't yet taken on an independent project that he could feel that he had ownership of, says Chein. “The LAURA provided him that chance, and he really grew through it.”
Today, Barone is a first-year PhD student at the University of Colorado Boulder. Both he and Dr. Chein credit this LAURA project as a launching pad for Barone’s current academic and professional pursuits.
“He probably would have gone to grad school anyway, but this gave him direction. It gave him something to really latch onto in his application process.”
Barone agrees:
“I think, without this LAURA project, I wouldn’t have really pushed to go for my PhD in Cognitive Psychology. The process gave me the time and resources to truly understand the scientific process.”