AI Tutoring in K-12 Classrooms: What's Actually Working in 2026
A look at real adoption data and early outcomes from AI tutoring rollouts, beyond the vendor hype.
AI Tutoring in K-12 Classrooms: What's Actually Working in 2026
Artificial intelligence has moved from pilot programs to daily routines in hundreds of U.S. school districts. Parents researching schools today are encountering a new question: how is AI being used in the classroom, and does it actually help kids learn?
The short answer is that AI tutoring is expanding rapidly across K-12, but evidence of effectiveness remains uneven. In this post, we'll walk through what the adoption numbers look like, where early results show promise, what challenges districts are facing, and what parents should ask when evaluating a school's approach to AI.
The Adoption Surge: From Pilots to Scale
The numbers tell a story of explosive growth. Khan Academy's Khanmigo, an AI-powered tutoring tool, jumped from 40,000 users to 700,000 students between the 2023-24 and 2024-25 school years, according to Khan Academy's chief learning officer Kristen DiCerbo. That figure is expected to exceed 1 million students in 2025-26. DiCerbo, who has worked in ed tech for two decades, called it the biggest one-year jump in education technology adoption she has ever seen.
Beyond Khanmigo, state-level initiatives are multiplying. Connecticut launched an AI pilot program in seven districts in spring 2025, introducing grades 7-12 students to state-approved AI tools with accompanying professional development for teachers. Indiana's AI-Powered Platform Pilot Grant funded one-year implementations in the 2023-24 school year, covering subscription fees and professional development to support high-dosage tutoring. Iowa committed $3 million to provide all elementary schools with an AI reading tutor that uses voice recognition, with rollout beginning in summer 2025.
These aren't niche experiments. As of March 2025, 28 states had published or adopted AI guidance for K-12 education. A September 2025 RAND survey found 54 percent of K-12 students reported using AI for school, up more than 15 percentage points in two years. On the educator side, student use of AI for school-related purposes jumped 26 percent since last school year, while educator use rose 21 percent over the same period.
The rollout is no longer hypothetical. AI tutoring is in classrooms right now, and the question is shifting from whether schools will adopt these tools to how they'll use them.
What the Evidence Shows
The hard truth is that rigorous evidence is still scarce. Khan Academy has not yet conducted a randomized controlled trial of Khanmigo due to the expense and logistical challenges, though the nonprofit plans to pursue that gold-standard research. Much of what we know comes from early-stage pilots, surveys, and correlational studies.
Still, there are promising signals. A 2026 pilot involving 15,000 students across 200 schools found that students who engaged with Khanmigo for at least 30 minutes weekly showed learning gains equivalent to an additional 2-3 weeks of traditional instruction. Khan Academy's broader efficacy research suggests that students who use their platform at the recommended level see about 20 percent higher-than-expected gains on state tests.
Outside the U.S., results vary. A 2020 study comparing Squirrel AI to human teachers found the AI system produced learning gains comparable to experienced human tutors, with students averaging 5.4 times faster learning in certain math topics. A 2025 RAND Corporation study of Carnegie Learning's MATHia showed effect sizes ranging from 0.19 to 0.36 standard deviations, indicating measurable gains in math achievement.
Reading tools also show potential. One randomized controlled study in 2025 found positive effects on elementary and middle school students' reading scores from the Dysolve program, which targets phonemic awareness for students with reading disabilities. Studies of Chinese kindergarteners using AI chatbots showed gains in language development, vocabulary, and syntax, though students who read with parents had better listening comprehension outcomes.
In Oklahoma, one high school participating in a Khanmigo pilot reported zero failing students in geometry after one semester of use, though school officials noted they are still analyzing the broader performance data.
The picture is more complicated in other locations. In Indiana, just over half of surveyed teachers reported that an AI tutoring platform had a positive or very positive impact on their teaching practice and student learning, while 40 percent reported no changes. In New Mexico, a text-messaging AI tool aimed at parent communication may have slightly increased attendance rates, but outdated student data and inconsistent parent engagement were barriers.
One Washington state alternative high school offers a more dramatic case study. Henderson Bay High School in Peninsula School District saw its ELA proficiency jump from under 40 percent to 66.7 percent in one year after implementing an AI integration pilot. The district credits the AI pilot heavily, though officials acknowledge other factors also contributed.
The takeaway: AI tutoring can produce learning gains, but results depend heavily on implementation quality, teacher training, and how consistently students engage with the tool.
Personalized Learning at Scale, With Caveats
The promise of AI tutoring rests on personalized learning. Unlike one-size-fits-all instruction, AI-powered platforms like Khanmigo adapt to each student's pace, identify specific gaps, and provide targeted practice. At Enid High School in Oklahoma, math teacher Stephanie Garis said Khanmigo helps her address individual students' algebra gaps, something she couldn't do effectively with 30 students in a classroom. The tool also personalizes examples based on student interests, framing math problems using sports or other contexts students care about.
At Nina Otero Community School in Santa Fe, New Mexico, teachers use Amira Learning, an AI reading tutor for grades K-6 that records students as they read and provides immediate feedback. If a student mispronounces a word or misunderstands context, an AI character helps them work through it. Teachers receive students' reading levels and tailored lesson suggestions, freeing up time to focus on students who are struggling most.
These examples highlight a key theme: AI tutoring works best when it augments, rather than replaces, human teachers. Research and pilot programs show the strongest gains when AI offers individualized feedback and tailored practice while educators focus on higher-order instruction and student connection.
But personalized learning through AI introduces new complexities. Fully personalized learning can isolate students, stripping away the collaboration and mentorship inherent in group settings. Even when AI tailors instruction, educators must carefully interpret its data, separating momentary struggles from true learning gaps to avoid misdirected interventions.
A Chalkbeat analysis noted that while AI advocates tout massive potential gains based on a famous 1984 study, those results have not been widely replicated. Modern studies of large-scale human tutoring programs show it is one of the most effective strategies for boosting learning, but results are about a tenth the size of what the 1984 study reported. AI is not as good as a human tutor, and setting unrealistically high expectations may lead to disappointment.
The Implementation Challenge
AI integration is not a plug-and-play solution. District leaders at the 2026 ASU+GSV conference shared critical mistakes they made when bringing AI into their schools. One common error: rolling out AI tools too quickly without gauging teacher readiness. Districts that initially tried to implement AI across all schools found it more effective to start with a smaller group of willing principals and teachers, using their experiences as a benchmark before expanding.
At Gem Prep, an Idaho charter school, leadership made the mistake of not giving AI training enough time. It's important not to make assumptions about adult readiness, said Chief Academic Officer Laurie Wolfe.
Teacher training is consistently cited as essential. In Oklahoma, teachers participating in a Khanmigo pilot received regular online training through Project ECHO before putting the tool into their classrooms. The training addressed common fears about AI, including whether students would use it to cheat. (Khanmigo includes guardrails to discourage cheating and help teachers identify plagiarism.) Tyler Elders, an instructional coach at Enid High School, called it the easiest implementation of a new technology he had ever done, crediting the ECHO training as absolutely essential.
A two-phase pilot across Michigan schools found that AI has potential, but only with intentional support. Teachers emphasized the importance of starting the year with an AI tool like Khanmigo rather than trying to integrate it midway through the semester. They also noted the need to set clear expectations and guidance on when and why to use the tool. One teacher observed that students don't naturally want to use AI the way educators hope they will, and that appropriate usage must be modeled consistently.
A systematic review of 43 empirical studies on AI integration in K-12 found that technical training alone is not sufficient. Successful integration requires a combination of pedagogical knowledge, positive attitudes, organizational support, and continuous training.
Equity is another major concern. Only about half of school districts offer AI training, leaving many educators feeling unequipped to teach AI literacy. The most pressing challenge ahead is ensuring that the benefits of AI in education reach students in low-income, rural, and under-resourced communities at the same rate as those in well-funded institutions.
What Parents Should Ask
If you're researching schools and AI tutoring comes up, here are concrete questions to ask:
What specific AI tools does the school use, and for what purpose? Look for schools that can explain the tool's function clearly, whether it's reading support, math practice, or writing feedback.
How much time do students spend with AI tutoring each week? Research suggests consistency matters. Students who use tools like Khanmigo for at least 30 minutes weekly tend to see measurable gains.
What training have teachers received? Effective AI implementation depends on teacher readiness. Schools that invest in professional development tend to see better outcomes.
How does the school balance AI with human instruction? AI should complement, not replace, teachers. Schools that position AI as a tool for practice and feedback while teachers focus on higher-order thinking and mentorship are on the right track.
What guardrails are in place? Ask about data privacy, how student interactions are monitored, and whether the tool includes safeguards against over-reliance or inappropriate use.
Are there any early results? Some schools track usage data, survey teachers and students, or monitor changes in test scores or grades. While rigorous evidence is still emerging, schools that are actively evaluating their AI tools are more likely to course-correct when something isn't working.
The Bottom Line for 2026
AI tutoring in K-12 is no longer experimental. Hundreds of thousands of students are using these tools daily, and adoption is accelerating. Early evidence suggests AI can produce learning gains, especially when students use it consistently and teachers are well-trained. But the technology is not a silver bullet, and results vary widely depending on implementation.
For parents, the key is to look past the hype and ask specific questions about how a school is using AI, how teachers are supported, and what early results look like. Schools that approach AI thoughtfully, with clear goals and ongoing evaluation, are more likely to deliver meaningful benefits for students. Those that rush into adoption without adequate training or infrastructure may struggle.
The AI tutoring story in K-12 is still being written. What happens next depends on whether schools can match the pace of technological change with the human work of training teachers, supporting students, and ensuring equitable access.