How AI is helping museum exhibits to teach
by Ben Kuchera
Illustrations by Jon McCormack
How do you make sure that every child who visits a museum has the same opportunities to learn as other children do? How can we increase the chances that each kid can truly comprehend exhibits without raising the pressure for museum staff to station a qualified instructor at each interactive station? The answer may be cameras, computers, and a virtual teacher who is able to work with children — not just lecture at them.
Jon McCormack
“Current exhibits rely on parents, signage, and staff/facilitators in the museums and science centers to provide support and guidance,” researchers at Carnegie Mellon University's Human-Computer Interaction Institute (HCII) explain. “Since it’s not always feasible to have knowledgeable staff and not all parents have the same knowledge, children do not get the same support.”
The group experimented with a combination of sensing technology and custom software to improve how museum exhibits teach scientific concepts to visitors by reacting to the actions of each visitor. This new thinking behind education in museums isn’t just about explaining the lesson in each experiment. It’s also about tracking the actions of the visitor and providing feedback in real time, without the need for human instruction.
“To do so, we developed new AI vision technology and Intelligent Science Exhibits that allows the system to observe students’ actions — to accurately monitor and evaluate predictions, experiments, and explanations and provide an intelligent guided learning experience,” the paper explained.
In this case, an exhibit was created in which children placed different towers on a table that simulated an earthquake while a computer-controlled character explained the experiment. Then, by tracking which towers were being used and when they fell during the simulated earthquake, the virtual character could ask questions about why one tower fell and another stayed up longer or offer suggestions about other strategies to try.
But how do you determine whether these systems are superior to previous methods for teaching science in museums? Does introducing a computer-controlled “teacher” into the mix decrease the self-directed play and learning that make museums so special?
The researchers found that visitors were spending more time at the exhibit with the AI instruction and were learning more about the science behind the experiments compared to the traditional design.
Jon McCormack
“Our results demonstrate children receiving interactive guidance and feedback while predicting and explaining are better able to learn to apply science in engineering tasks, especially when combined with exploration and construction activities,” the researchers wrote.
Balancing self-guided play and experimentation with learning has always been a challenge for museums, but this experiment shows a way forward that increases participation in learning, doesn’t rely on the parent to understand and teach the lesson, and doesn’t require museums to staff the exhibit throughout the day. It’s able to engage with the children, instead of lecturing at them, and it sounds as if that shift makes all the difference.
As someone who loves museums and is always trying to get their kids to spend more time reading plaques to get the whole story behind each display, this also just seems a lot more fun.