“I wasn’t always able to get great feedback to the students,” he said. “I would try to get to it as much as I could, but a lot of those would just be the students receiving a grade without any feedback.”
When Kiddom, the classroom management system used in Providence Public Schools where Hamill worked, announced a beta version of an artificial intelligence-powered automated feedback feature for math teachers, he was ready to give it a try. The beta testing took place for about four months during the 2023-24 school year.
“The thing I noticed most was that it was a huge quality-of-life upgrade for us as teachers. I used it mostly for end-of-day exit tickets, to expedite that process of creating and getting feedback to the students, and it was a giant help,” Hamill said. “It allowed me more time to focus on lessons and, honestly, just to be well rested and ready to go in the morning.”
The AI tool was able to give every student the kind of personalized responses that can help improve learning, he said, which he did not have time to provide manually. Rhode Island’s core educational standards are built into the Kiddom system, Hamill said, so the AI feedback tool was able to tie any mistakes to specific standards for further review.
“It would recognize where the mistake was made, and it would just be like, ‘Hey, refresh information from this standard,’ and then it would direct them,” he said. “So, it might not say exactly what they did wrong, but it pushed them in the right direction to recheck what they could study more on to help get the question right.”
Along with the automated feedback feature, Hamill was given access to three other new AI tools for math teachers: automated grading, a practice problem generator and a lesson clipper to prepare plans for substitutes. For the AI-driven grading and feedback tools, he said teachers must review each score or response, then click to accept, decline or edit what the AI has generated.
While the automated feedback tool had the biggest impact on his own teaching practice, Hamill said the other three will save teachers time as well. However, the AI grading was not as "polished" as the feedback, he said, marking answers incorrectly about 15 to 20 percent of the time.
“I was a bigger fan of the feedback, because there was nothing wrong with the feedback,” he said. “The scoring, it would maybe not recognize fractions right in different ways or something like that, so there were a few small things that could be ironed out.”
Kiddom officially rolled out three of the four AI tools within its math curriculum offerings this month, with the lesson clipper for substitutes “coming soon,” according to the company’s website.