Khan Academy is launching a redesigned platform that embeds its artificial intelligence assistant, Khanmigo, directly into student practice sessions. The update follows a year-long pilot program involving over 5,000 students, which revealed that standalone chatbot interfaces often disrupt classroom learning.
What Happened
According to a recent announcement by Khan Academy, the educational nonprofit rebuilt its platform after realizing that students rarely paused their work to open a separate AI chat window for help. During the 2025–26 school year, the organization piloted this integrated system with 5,100 students across five U.S. school districts, including Hanover Community School Corporation in Indiana and Taft ISD in rural Texas.
The new design changes how the AI interacts with struggling students. Previously, Khanmigo adhered to strict guardrails that prevented it from giving direct answers, forcing students to struggle productively. However, district leaders noted that when students completed homework incorrectly, it took seven times the effort to unlearn those mistakes. In a recent webinar, Hanover Assistant Superintendent Phil Misecko explained that the updated AI acts as an active safety net, walking students through errors immediately so they do not practice math incorrectly. Now, the AI offers gentle hints before a student submits an answer, but provides more direct, step-by-step guidance after an incorrect attempt.
The Bigger Picture
While Khan Academy aims to improve academic performance on state exams, independent evidence linking AI tutors to higher standardized test scores remains limited. Currently, the organization evaluates Khanmigo using internal metrics. According to the Khan Academy Blog, developers track "next-item correctness" to see if a student can solve the very next problem without assistance, alongside "cognitive engagement quality" evaluated through the GrowthBook A/B testing platform. Some independent reviews, such as an eight-week test by Kids AI Tools, showed double-digit percentage improvements on math and science concept tests, but these involved small sample sizes of only 12 students.
Educational researchers also warn about the dangers of over-corrective AI. A study from the Wharton School's Mack Institute found that generative AI tools can actually harm long-term learning if they make help too easily accessible, which leads to an over-reliance that prevents independent problem-solving. This tension is central to math instruction; while a systematic review in Frontiers shows that adaptive AI feedback improves problem-solving, it also cautions that too many hints restrict the development of independent skills.
The success of the platform's new teacher dashboards depends heavily on school environments. While research published by the University of Virginia demonstrates that classroom dashboards significantly improve a teacher's ability to notice and respond to student struggles, a study by the Arizona Board of Regents cautions that actual classroom adoption is closely tied to school funding, teacher confidence, and "technostress."
What This Means for Families
As we previously reported, families are increasingly turning to AI tools for homework help, but the way these tools are designed matters. The integration of AI directly into practice screens means students will receive automated support without having to seek it out. For parents, this can reduce homework frustration. However, because immediate feedback can sometimes act as a cognitive crutch, parents must ensure that children can still solve problems without an AI prompting them at every step. This shift also places new demands on schools, which must balance these tools with teacher training and centralized data management.
What You Can Do
Parents can take several steps to help their children adjust to these changes. First, monitor the level of support by asking your child to occasionally solve math problems on paper without the AI active. This helps make sure they can perform independent problem-solving without digital hints.
Second, talk to teachers about dashboard data. Ask your child’s teacher how they use the new Khan Academy dashboard insights to target small-group instruction or adjust classroom lessons.
Finally, encourage active reflection. When the AI explains a mistake, have your child explain the correction back to you in their own words to solidify conceptual understanding.