In this way, AI is modeling the dialogue between teachers and students. The tools adapt to the students’ input, progress, and results and shape the process in the way that a teacher would—helping the students to correct their mistakes and learn new things.
With regard to math, AI is helping young students to “think math” by exploring the concepts behind the math through modeling, manipulatives, and multiple representations (such as base-10 blocks and place value cards). With the addition of real-time, adaptive feedback and scaffolded instruction, the application is learning about each student, too.
AI in action
Take, for example, Slackwood Elementary School in Lawrence Township, New Jersey. George Regan, a math specialist for the district, observed that apps driven by AI helped students build a solid foundation in conceptual understanding and had a positive effect on instilling a growth mindset among the students.
What’s more, using AI-embedded applications improved students’ exam scores: At the end of the year, almost 100 percent of first-grade students achieved a benchmark score of 35; at the beginning of the year, 60 percent of those students achieved a benchmark score of 9, indicating that about ⅔ of the students had demonstrated readiness.
Regan had started using an AI-driven personal teacher assistant, Happy Numbers, for 25 percent of the students’ math time; it served as an independent math station. Typically, in a math block of 60 minutes, four or five stations rotate every 15 to 20 minutes, with an AI tool used for small-group instruction at one of the stations.
Shortly after seeing the first-graders’ success with the exams, he decided to start applying this approach to kindergarteners, too. He gave end-of-the-year kindergarteners the same benchmark assessment that he had given the beginning-of-the-year first-graders. Remarkably, the kindergarteners scored 90 percent on this exam.
Most teachers are not surprised that a certain level of anxiety often comes with math learning. Noticing a change in student behavior, Regan observed that the kindergarteners demonstrated little to no anxiety during the benchmark exam. They were correctly answering abstract equations, having successfully internalized the models due to strong conceptual understanding. In other words, they were no longer just memorizing answers to rote facts; instead, they were becoming fact fluent by understanding the concepts behind—and making sense of—the numbers.
While the end-of-year kindergarteners achieved 90 percent on a benchmark math exam designed for beginning-of-the-year first-graders, the first graders had achieved 60 percent on this same exam when they had started that year. Regan wanted to know whether the first-graders’ scores were a byproduct of summer slide, if there was simply a difference in the two cohorts of students, or if there was more to the story. He decided to collect more data and gave the new class of first-graders (who were the same end-of-the-year kindergarteners) the exam again. After achieving 90 percent at the end of kindergarten, they achieved 88 percent at the start of first grade. This result was much higher than the 60 percent achieved by the beginning-of-the-year first graders the year before.
Regan felt that the student growth and improvement in higher-order thinking could be attributed to true learning, and not to erroneously low September scores.
Creating personalized learning is difficult to do at scale; providing a live tutor for each student is cost prohibitive at best. However, thanks to comprehensive and developmentally appropriate AI technology tools that integrate into and assist with small-group instruction, schools can see the results they seek: independent student success and achievement.
- 3 ways to bring teacher PD into the 21st century - December 5, 2022
- Research points to a widening academic divide - December 5, 2022
- 6 key elements to build a successful coaching program - December 2, 2022