AI-infused teaching that really helps kids learn? Schools and learning hubs are embracing tech made possible by the National Science Foundation

Parenting has never been an easy task. But in the era of artificial intelligence, it can feel even more complicated. Algorithms, chatbots and smart devices are becoming more deeply threaded through our lives — and our children’s lives — with each passing day. We know that as AI’s rise continues, this evolving technology is going to play an ever-growing role in the world our kids inherit.

So it’s clear that trying to avoid AI or keep America’s kids away from it isn’t possible, and it’s certainly not a winning strategy. It’s vital that today’s students become fluent in interacting with AI and understanding the many ways it’s evolving.

Schools, after-school learning programs and the places we take our kids — like science centers and children’s museums — need to introduce AI in ways that serve young learners well. That means using AI in ways that support the efforts of teachers and parents, rather than attempting to replace them, and ensuring that kids aren’t simply passive consumers of information.

A groundbreaking, patented project based in Pittsburgh at Carnegie Mellon University is currently working to do just that. With some help from an animated gorilla, the tech experts at NoRILLA are facilitating the work of teachers and museum educators while increasing kids’ interest in hands-on science learning and helping them grasp scientific concepts from a young age.

21ST-CENTURY LEARNING SPARKED BY A 20TH-CENTURY ICON
Many families may be including a visit to a science center or children’s museum in their summer plans. If they’re lucky they’ll come upon NoRILLA Intelligent Science Exhibits — AI-fueled learning stations created by Dr. Nesra Yannier and her team at Carnegie Mellon.

Designed in partnership with Pittsburgh’s Carnegie Science Center, the Children’s Museum of Atlanta, Philadelphia’s Please Touch Museum and the Children’s Discovery Museum of San Jose, NoRilla’s high-tech EarthShake grew from a wonderfully low-tech source: A grant and award Yannier received as a recipient of the Fred Rogers Memorial Scholarship in New Media for Children, given by the Television Academy as part of the Emmy’s College Television Awards in 2013.

Growing up in Turkey, Yannier didn’t know much about “Mister Rogers’ Neighborhood.” But as a graduate student at Carnegie Mellon University, she began learning about this man that everyone in Pittsburgh seemed to adore. She soon discovered “his values and everything he did to improve children’s lives,” she says. “It really resonated with my goals, and what I wanted to do.”

“The Fred Rogers award gave me the funding to take this work out of the laboratory and begin to test it out in a museum setting. That was the first catalyst,” she said. The vital next step, though, was funding from the National Science Foundation (NSF), where Yannier is a Principal Investigator (PI) on this project, along with Professor Ken Koedinger and Professor Scott Hudson.

“The NSF grants allowed us to build on it,” Yannier says, and that’s how an Intelligent Science Exhibit called EarthShake and additional mixed-reality learning resources have since expanded to more than 35 locations including children’s museums and science centers around the country and in Europe.

Without this funding, Yannier could not have done the research to blend AI technology with in-person, hands-on learning and “bring it out into the real world to make it beneficial for children,” she says.

“A lot of museums right now are looking into ways to incorporate AI, but they don’t want to incorporate technology just for the sake of technology,” Yannier explains. “So having something that actually has the proven learning benefits and the outcomes is really important.”

An initial NSF grant built on Yannier’s early work and allowed her team to collaborate with the four museums and fine-tune the project to make sure students’ science learning and interest was measurably increasing. A second NSF grant, she says, was vital because it “gave us a way of transitioning what we’ve done in the museums into school classrooms.”

HOW DOES IT ACTUALLY WORK?
NoRILLA is designed to foster curiosity, collaboration and persistence – and it puts real-world, hands-on learning in the foreground, literally.

Here’s how it works: A student stands at a table doing hands-on, physical experimentation – building with blocks, for example. An AI-powered computer visioning system can see and understand what the child is doing in the physical world. There is a TV screen in the background, behind the table, and on that screen an animated gorilla “will ask the student to perform different physical experiments on the table,” Yannier says.

“In our first intelligent exhibit, EarthShake, the system will detect whether they place the correct structure or not,” she says, “and will ask them to make a prediction about which of these structures would fall first, if the table shakes.”

When the table does shake, AI detects which structure fell. With that information, the gorilla will give the student personal feedback and ask questions about their particular prediction, helping them understand what happened from a scientific perspective.

“Once they learn the scientific principles, it’ll give them different challenges, like, could you make a structure that is taller than the Eiffel Tower, or that has a thinner base? And then the AI can check whatever structure they build, and give them hints and suggestions about how they can make it better,” Yannier explains. “Maybe if it’s not symmetrical, it’ll give them feedback on that. Or if it has more weight on the top, it helps them understand why that makes it less stable. In real time, it can detect whatever structure they build with different materials, and then it will give them real time feedback based on that. It also adapts based on their background and gives guidance and feedback accordingly.”

The NoRILLA team has started expanding the mixed-reality AI system to many different content areas. They have also created Smart-Ramps, where the student does experiments using physical ramps on the table. AI tracks the objects the kids build or place on the ramps, as the kids make predictions about which object will go faster. Again, this new module on the NoRILLA platform combines the joy of hands-on exploration while using AI to scaffold deeper thinking.

“They’re learning about potential energy, gravity, friction and other STEAM concepts that involve important higher-level thinking skills,” Yannier says, while “the gorilla character gives them personalized feedback and guidance as they go through the scientific process.”

MEASURING THE IMPACT
Extensive research has shown that it’s working.

“A lot of times in the museums, kids go from one exhibit to the other, and don’t spend much time. We wanted to increase both the engagement and the learning,” Yannier says.

The NoRILLA team did a series of experiments to understand whether having this interactive AI layer on top of the physical experimentation improved kids’ learning. Would their understanding of what’s going on transfer to real world problem-solving skills? Would it increase how much time they spent at each exhibit?

“We compared our intelligent science exhibit with the same exhibit that had the AI turned off. The version without the AI was more comparable to the museum exhibits that are out there currently — the standard exhibits. We saw that when you added the AI layer on top, kids were able to understand the scientific principles significantly better. They were able to explain why things were happening much more clearly,” she says.

“But what was more interesting was that the students also could use what they learned and transfer that to engineering and problem-solving skills. We did pre-tests and post-tests to see that. In the version without AI, kids were doing a lot of building but their buildings did not improve over time at all. When we added the AI layer, their buildings and their problem solving skills improved by approximately 10 times more.”

Kids also spent four times as many minutes engaging with the lessons at the AI-powered exhibits, compared to the same exhibit without AI. And rather than eliminating the need for human teachers, these hands-on lessons can actually enhance the work of teachers and parents.

SUPPORTING TEACHERS AND PARENTS
“In our research, we also saw that having that AI layer was also improving the scientific conversations between the parents and the kids. The parents would start asking more questions to the kids, mimicking the gorilla character in the system, and they would have deeper scientific conversations, as well,” Yannier says. “That’s an ongoing part of our research, because at the museums, we also want families and kids to interact with each other. With the AI layer included, we see that even if the parent has no background in science, the system guides the parent to help the kids, which is nice.”

Teachers who encountered NoRILLA on school field trips also interacted more deeply with students and they began asking Yannier and her team if NoRILLA could be adapted for classroom use.

What if the library or maker space at an elementary school had a station equipped with NoRilla technology, fueled by lessons that were designed for each grade level? Could it be proven to be helpful for students and teachers?

Funding from the NSF made it possible to begin piloting NoRILLA in four Pittsburgh-region school districts — Montour, Avonworth, Elizabeth Forward and Clairton City.

“We wanted to understand teachers’ needs, so we started doing classroom observation,” Yannier says. “We found that part of the problem in many communities is that the elementary teachers really want to teach science, but a lot of them don’t have a science background. They really liked this idea of something that could support them together with the kids, because it would answer the kids’ questions and guide the students and the teachers together.”

Teachers asked for a curriculum using NoRILLA tech that aligns with the lessons and the state standards that they’re teaching. With help from the NSF funding, Yannier says, “we started creating curriculum and lesson plans lesson plans that are aligned with the national and state standards, working closely with teachers to make it easier to integrate it into their everyday classroom experiences.”

Just like in the museums and science centers, the data was impressive.

“We’ve actually seen very high learning gains in the classroom: the scientific explanations that show their deeper, critical thinking skills that are aligned with the Next Generation Science Standards have tripled in three months,” Yannier says.

“There were really high learning gains, and also very high engagement gains, both for girls and boys. So that has been really exciting. We are working on partnering with many schools and museums around the U.S. and internationally to make this system available and beneficial to as many more children as possible.”

For parents and teachers, learning tools like NoRilla provide a meaningful way to get students interacting with AI while still keeping human interaction and face-to-face connection at the forefront.

It’s an embrace of AI that encourages human interaction rather than limiting it, Yannier says, and helps today’s students grasp the science and technology skills that can help them thrive in our tech-infused world — no matter what the future brings.