By: Dr. Kim Abel // Edited By: Dr. Reed Randall
A new conversation is happening in schools right now. With the arrival of adaptive AI that coaches students toward answers, teachers are asking what they are measuring and what is now relevant to measure.
Consider a 9th grader who solves a quadratic equation. The answer is correct. With AI a prompt away, the correct answer has become the starting point anyone can reach, rather than a measure of understanding.
For over a century, schools have measured what students know at the end of learning. AI is changing that architecture. For the first time, educators can measure the learning process itself in real time, watching where understanding forms, where it breaks down, and how students grow when challenged at exactly the right moment. That shift raises a question our teachers at Optima Academy Online began asking early: if every student can reach the correct answer with AI a prompt away, what does a grade actually mean anymore?
“The real measure of learning is no longer the final answer. It is the process of understanding.”
— Optima Academy Online, AI School Policy
What the data is already showing us
Mrs. Arreola teaches 4th grade math at Optima Academy Online. Like a typical classroom, her students span a wide range of math abilities, from Kindergarten to 6th grade. Midway through the 2025-26 school year, she introduced an AI tool that changed how her students experienced learning. The tool invited students to ask questions about their assignments, engaged them in mini-lessons targeting areas of uncertainty, and guided them toward filling in the conceptual gaps the tool identified along the way.
Mrs. Arreola quickly learned that right or wrong answers were not the whole story. She tracked students’ engagement depth, watching how they moved through daily assignments using the AI agent. After each assignment, the AI system sent her a color-coded report showing exactly where each student stood in their grasp of the mathematical process.
🟢 Arrived at the answer unassisted
🟡 Needed some assistance
🔴 Needed significant help or did not understand the concept
The AI system also tracked time on task for each student. The time-on-task data told an important story. Students who engaged with the tool consistently throughout their assignments, not just when they were stuck at the end, showed dramatically more growth than those who reached for AI assistance only at the finish line.
Norm-referenced test data confirmed what Mrs. Arreola was already seeing. At the start of the year, 14% of her students were performing two or more grade levels below 4th grade. By the end, that number had dropped to just 1%. What made the results truly remarkable was what happened at the other end of the student performance spectrum. In many classrooms, an intense focus on struggling students comes at a cost to those who are ready to move ahead. In this case, advanced students also grew. The percentage of students performing above grade level grew from 14% to 27%, according to iReady assessment data. Mrs. Arreola attributed that growth across the board to adaptive AI that met every student exactly where they were.
A landmark study published in PNAS in 2025 sheds light on what made the difference in Mrs. Arreola’s classroom. Researchers conducted a large-scale randomized controlled trial with nearly 1,000 high school math students and found that when students used AI that simply gave them answers, they performed worse on subsequent assessments taken without AI assistance, a 17% reduction compared to students who had no AI access at all. [1] The difference at Optima Academy Online was in how the AI was used. Students were guided through their thinking, challenged at the point of their mistakes, and required to arrive at understanding on their own terms.
“I’ve noticed that the ones who use it frequently to complete assignments have made so much more growth. I like that they have a sort of teacher guidance as they make a mistake, which is much more meaningful than feedback after they have already finished. It helps them understand right in that moment.”
— Mrs. Tina Arreola, 4th grade teacher, Optima Academy Online
One student, Emily, became the clearest case study. She had struggled at the start of the year, but as a consistent user of the AI tool both in and outside of class, something shifted. By spring, her confidence in math had grown “leaps and bounds,” and she had become a different learner. For Emily, the AI became a thinking partner. What changed for her was the process itself, the guided struggle, the feedback she received in the moment, and the adjustment in her thinking that her AI agent made possible. This kind of in-the-moment, corrective feedback is central to what researchers have long identified as the most powerful driver of student growth. A 2024 study on learning from errors found that when students receive feedback at the point of a mistake rather than after the fact, the experience of working through that error becomes a meaningful learning event in itself. [2] Working through the process made Emily a better thinker, which is exactly what Mrs. Arreola had planned.
With the AI handling the detailed work of identifying gaps and guiding students through them, Mrs. Arreola found a new kind of freedom in her classroom. She focused on the human side of student development, designing lessons around curiosity, discovery, and genuine interest in mathematics. Her energy went toward making math feel full of wonder. This is what well-used AI makes possible in a classroom where the irreplaceable human work of inspiring, connecting, and igniting curiosity finally has room to flourish.
This kind of growth reflects a partnership with a tool and a teacher. At Optima Academy Online, AI works alongside teacher mentorship, immersive virtual environments, and personalized learning pathways built around each student’s readiness. Mrs. Arreola’s classroom data reflects how these layers work together.
How high school math teachers are grading the process
High school math teachers Mrs. Silva and Ms. Loertscher created grading rubrics to make a student’s thought process the primary object of assessment. Accuracy and following directions are still part of the grade, but the majority of the grading score comes from a verbal response that demonstrates understanding and reasoning, clear steps and organization, and an authentic mathematical voice using proper math language. Each teacher developed rubrics specific to her subject area, covering Algebra, Geometry, and Pre-Calculus.
“Students who could not answer simple questions about process or reasoning in class were turning in objectively perfect work. There was a massive discrepancy between what I was seeing in class and what was being submitted.”
— Ms. Megan Loertscher, High School Math Teacher, Optima Academy Online
These grading rubrics were built around a simple truth. Genuine understanding cannot be borrowed from a machine. A student can use AI to arrive at the right answer, but submitting their reasoning in their own authentic mathematical voice is entirely a different matter. Teachers can hear the disconnect between an explanation and the actual work shown. The rubric makes that gap visible, and more importantly, it makes genuine understanding the only path to a strong grade.
“This is especially valuable in math, where I can see whether students truly understand the process and reasoning behind a problem. The rubric reveals their thinking in a way a final answer alone never could.”
— Mrs. Savanna Silva, High School Math Teacher, Optima Academy Online
Cognitive science has long established why verbalizing student understanding matters. Research consistently finds that students who engage in self-explanation, articulating their reasoning in their own words, perform significantly better on assessments than those who do not. [3] When Mrs. Silva and Ms. Loertscher ask students to demonstrate an authentic mathematical voice, they are requiring the cognitive act that builds genuine understanding.
“I wanted something that would allow me to award students who are explaining their work with their own reasoning and in their own words, regardless of the correctness. That is how the Authentic Voice category was born.”
— Ms. Megan Loertscher, High School Math Teacher, Optima Academy Online
The shift from score to conversation
The most important reframe of the grading process is philosophical rather than technical. Assessment in an AI-integrated classroom is an ongoing conversation, present from the first attempt to the final answer. In practice, this means students interact with AI agents that ask follow-up questions, probe understanding, and push students to explain their reasoning rather than simply confirming whether an answer is correct.
This shift asks educators to reframe what counts as evidence of learning. Grading the journey means weighing process, iteration, and reasoning at least as heavily as accuracy, because growth is gradeable in ways a final score never captures. At Optima Academy Online, making the thought process visible looks like students filming themselves speaking in a learning management system, joining a teacher on camera to talk through a math problem in their own words, or recording their avatar in a virtual environment explaining their reasoning step by step. Students must use mathematical terminology appropriately and in their own voice. Either they understand the process, or the camera shows exactly where the thinking breaks down.
Tracking engagement over time adds another dimension to assessment. Teachers watch which students interact with AI tools consistently throughout their assignments and which ones reach for help only at the finish line. That pattern tells you who is building understanding and who is bypassing it.
Grading rubrics should include room for iteration and room for authentic voice. Iteration means scoring first attempts, second attempts, and the distance traveled between them. Authentic voice means a student who explains their thinking imperfectly, in their own words, and with their own stumbles is demonstrating something a polished AI-generated response never could. Research comparing AI systems designed to give answers versus those designed to guide students with hints found that the hint-driven approach produced qualitatively richer student interactions, with students asking for clarifications, interpreting guidance, and reconsidering their strategies. [4] The design of the AI matters as much as its presence.
“The ones who struggled and kept going showed more growth than the ones who always got it right.”
— Mrs. Tina Arreola, reflecting on a full year of AI-assisted learning data
What this means for the future of assessment
The implications of changing grade focus reach beyond any single classroom. AI that tracks the learning process in real time exposes a gap in the traditional report card design. A letter grade tells a family whether their child arrived at the right answer. It tells them almost nothing about how many times they tried, how they responded when they were wrong, or what they genuinely understand.
The schools leading in this next era are asking harder questions now. What if report cards showed growth trajectories alongside grades? What if AI-generated learning profiles gave teachers, students, and families a richer picture of mastery than any single score could provide?
The meaning of grades is being renegotiated, and the schools participating in that conversation now will shape what learning looks like for the next generation.
Rethinking mastery
Mastery in the AI era is not about producing a correct answer. Any student with a phone can look up an answer. Mastery is demonstrating that you understand why the answer is correct, that you can explain it, adapt it, apply it in a new context, catch your own mistakes, and keep going when you are wrong. These are deeply human skills, and they have always been the true measure of learning.
Mastery built on understanding, iteration, and persistence can still be graded. It just requires rubrics built for the world we are actually teaching in, where teachers are free to focus on guiding students through real application, sparking discovery, and honoring the curiosity that drives genuine understanding.
Emily’s story is ultimately about what happens when a student is seen as a learner first. When she struggled, she was supported. When she iterated, she grew. The journey of becoming someone who truly understands is what a grade should reflect.
So what does a grade mean anymore? It means something better than it ever did. It means we finally have the tools to measure what great teachers have always known matters most: not the answer a student found, but the thinker they are becoming.
References
[1] Bastani, H. et al. (2025). Generative AI without guardrails can harm learning. Proceedings of the National Academy of Sciences. Nearly 1,000 high school students who used answer-giving AI scored 17% worse on follow-up assessments than students with no AI access at all. Pedagogical design, not technical sophistication, was the determining variable.
[2] Metcalfe, J., Xu, J., Vuorre, M., & Bjork, R. A. (2024). Learning from errors versus explicit instruction in preparation for a test that counts. Journal of Experimental Psychology. Students who worked through errors at the moment they occurred retained learning more durably than those who received correct information directly.
[3] Chi, M. T. H., De Leeuw, N., Chiu, M., & LaVancher, C. (1994). Eliciting self-explanations improves understanding. Cognitive Science, 18, 439-477. Students who articulated their own reasoning during learning consistently outperformed those who simply reviewed correct solutions, across both mathematics and science contexts.
[4] Italian Economic Journal / Springer Nature (2025). Artificial intelligence in education: Computer-assisted learning and AI-guided tutors. Students using hint-driven AI engaged in richer learning interactions than those using answer-giving AI, asking more questions, interpreting guidance, and reconsidering their strategies.
