AI Shown to Increase Learning and Engagement in Science Museums

A team of researchers at Carnegie Mellon University’s Human-Computer Interaction Institute (HCII) have demonstrated a new way of effectively increasing learning and engagement for museum exhibits. The team relied on artificial intelligence (AI) to develop new interactive and hands-on exhibits, including an intelligent, virtual assistant that interacts with visitors. 

Compared to the traditional one, the new intelligent exhibit was found to increase learning and the time individuals spent at the exhibit. 

The team’s findings were published in the Journal of the Learning Sciences. 

Turning Play Into Learning

Nesra Yannier is an HCII faculty member and head of the research project.

“Having artificial intelligence and computer vision turned the play into learning,” Yannier said. 

The team also consisted of Ken Koedinger and Scott Hudson from CMU, Kevin Crowley from the University of Pittsburgh, and Youngwook Do from the Georgia Institute of Technology. 

One of the popular exhibits in museums are earthquake exhibits, where kids build towers and try to secure them from falling when the table shakes. There are often signs placed around the exhibit to engage the kids regarding science, but evidence is not clear how effective they are. 

AI-Enhanced Earthquake Table

Yannier led the team of researchers to build an AI-enhanced earthquake table that had a camera, touchscreen, and large display. It also had NoRilla, which is an intelligent agent in the form of a virtual gorilla. NoRilla replaces the traditional signs and interacts with participants by taking them through different challenges and posing questions regarding the earthquake exhibits, such as why the towers did or didn’t’ fall. 

The team tested the intelligent earthquake exhibits at the Carnegie Science Center in Pittsburgh. Children who attend a summer camp at the center interacted with either the intelligent or traditional exhibits, and they completed pre- and post-tests and surveys. The team also observed visitors as they interacted with the exhibit.

The surveys found that the children learned significantly more from the intelligent exhibit, and they had the same amount of fun when compared to the traditional one. The AI exhibit also helped children better understand the scientific concepts, as well as improved their building and engineering skills. 

The results of the study also revealed that people spent around six minutes at the intelligent exhibit, compared to 90 seconds at the traditional one. 

Koedinger is a professor in HCII.

“What’s particularly impressive to me is how the system engages kids in doing real scientific experimentation and thinking,” Koedinger said. “The kids not only get it, they also have more fun than usual exhibits even though more thinking is required.” 

“Our exhibit automated the guidance and support that make hands-on physical experimentation a valuable learning experience,” Yannier added. “In museums, parents may not have the relevant knowledge to help their children, and staff may not always be available. Using AI and computer vision, we can offer this experience to more children of different backgrounds and at a wider scale.”

Yannier says that this new technology could also help students in the classroom and at home, not just in a museum.