All solutions created by Knowledge AI are rooted in proven educational research methodology. Our theoretical framework supports every feature that we offer in our KAIT@HOME, KAIT@SCHOOL, and KAIT TEST PREP products.
Mastery Learning Loop and Bloom’s 2-Sigma
When students are acquiring new knowledge and skills, the single most significant factor that will determine their success is the amount, timing, and quality of the feedback they receive.
The graph below, from Benjamin Bloom’s ‘2-sigma problem’ research in 1984, plots the impact on academic performance based on 3 cohorts of students: conventional, mastery learning, and tutorial. Students in the conventional learning control group were learning in a traditional classroom with a 30 to 1 student-to-teacher ratio. Students in the mastery learning experimental group also had a 30 to 1 student to teacher ratio, however, the students received individualized and frequent feedback from their teacher. These students outperformed the students in the conventional learning control group by 1-sigma, I.e., the resulting distribution is one standard deviation better. Finally, the students in the tutorial experimental group were taught one-on-one and were given the quality and frequent feedback. These students outperformed the students in the conventional learning control group by 2-sigma, or the resulting distribution is two standard deviations better.
Instead of waiting for an end of unit or even end-of-chapter test, KAIT products allow for a formative assessment loop to monitor student progress. Every time an assessment is given our program analyzes student behavior, utilizes cognitive and behavioral metrics (CBM) to accurately identify performance level, and then our AI algorithm automatically generates individualized interventions throughout the mastery learning process.
Memory and Retention
Drawing suggestions from the Ebbinghaus’ forgetting curve*, teachers can design their schedules for lessons, tests, and informal assessments for their students. Technology solutions offer the potential to create learning environments with these principles that can be a valuable aid to teachers and students.
The graph below depicts Ebbinghaus’s forgetting curve which portrays the decline of memory retention in time. This image also shows how information is lost over time when there is no attempt to retain it. However, the more one attempts to retrieve the information from memory in a scheduled and well-thought-out way, the longer the new information is retained.
Ebbinghaus’s Forgetting Curve
(How much of something do we forget each day?)
Typical Forgetting Curve for Newly Learned Information
Additionally, it’s been found that retention increases when the process of knowledge acquisition happens within an active versus passive environment. KAIT products are cross-platform, interactive, and individualized to ensure active engagement with content for maximum retention.
Learning Pyramid. Retention Rate.
One of the most persistent problems in current education systems stems from the use of test scores and GPA as primary determinants of understanding. Too often, students are classified in terms of their grades: A, B, or C students. The world is in a constant search for an alternative measurement as grading systems not only yield inaccurate results, but they are also a major deterrent in motivation. For this reason, Knowledge AI has created the Understanding Index TM: a highly sophisticated measurement of understanding using various time and biometric data we collect from students’ problem solving.
In deriving the Understanding Index, we used multivariate analysis, including thinking time, total solving time, number of pauses, stroke count, and many behavioral characteristics as variables. Then we converted such variables into an Understanding Index which can distinguish students who truly understand the material versus those who have superficial understanding. The Understanding Index is used to accurately diagnose the Zone of Proximal Development for each student, based on the type of questions, subtopics, and courses.
KAIT incorporates the Understanding Index in creating AI-based recommendations tailored to strengthen each student’s gaps in learning, according to their understanding level rather than raw test scores.
Our AI model is unlike any other applications of AI in learning because KAIT captures students’ understanding levels. Instead of inputting binary data related to correctness (limited to a 0 or a 1), KAIT is more granular in its approach. By inputting understanding index scores that range from 0.1 to 0.9, our score and knowledge map prediction are all that much more precise.
Our recommendations are also concept based, rather than problem based. This means that when a student gets a problem wrong (C1), instead of recommending the same type of problem, our engine gives students additional problems related to underlying concepts that must be mastered (C2 and C3) in order to successfully solve the given problem.
A real-world example would be: if a student gets a Pythagorean theorem problem incorrect (C1). Depending on the student, the engine will recommend problems related to solving quadratic formulas (C2) and taking square roots (C3). In this way, KAIT helps students truly understand concepts instead of implementing compensatory mechanisms that often include rote memorization.
Individualized Learning – Zone of Proximal Development
Even with a mastery learning loop, spaced repetition, and active learning for retention, the content itself must be at an appropriate level for each learner for them to advance in subject matter knowledge. Information that is too easy is already known, while information that is too difficult is ignored. To teach a student effectively, the curriculum must fall within a student’s Zone of Proximal Development (ZPD). KAIT’s formative assessment cycle and CBM continuously and automatically identifies every student’s ZPD and suggests the exact Conceptual Building Blocks needed to support growth.