This is the fourth academic year since ChatGPT was launched and more artificial intelligence tutoring tools are finding their way into classrooms, according to a K-12 Dive article.
The number of K-12 students using nonprofit Khan Academy’s Khanmigo tutoring tool jumped from 40,000 to 700,000 between the 2023-24 and 2024-25 school years, according to Kristen DiCerbo, Khan Academy’s chief learning officer. The number is expected to reach more than one million students in 2025-26.
As the use of Khanmigo and other AI tutoring tools boom, there is still little hard evidence if they can actually improve student outcomes.
School and district leaders should consider these three key questions:
1) What AI tutoring tools are available?
Beyond Khanmigo, AI tools specifically catering to K-12 include Amira, with a cartoon avatar serving as a reading tutor and giving step by step help to students while sessions are recorded so teachers evaluate problem areas. AI-powered tools EarlyBird, Bamboo Learning and Imagine Learning use similar assessment techniques.
Publicly available AI apps also embrace tutoring options anyone can use. OpenAI in late July launched a “Study Mode” in Chat GPT using interactive prompts to ask questions and guide students through a problem rather than providing a quick answer. Google has a rival mode for Gemini called “Guided Learning,” and Anthropic’s Claude now offers a tutoring tool for college students.
These new tools arrive as some teachers worry about students using AI tools to cheat. The Pew Research Center found earlier this year students are increasingly using ChatGPT for their schoolwork, with 18% of surveyed teens saying it’s acceptable to tap into AI for assistance with essays.
2) How effective are tools?
Research has long backed human tutoring as an effective way to raise student achievement, says Robbie Torney, senior director of AI programs at Common Sense Media, a nonprofit research organization.
But there is little to no evidence that generative AI-powered tutors can do the same, Torney says.
There is a strong sense of urgency to help educators understand the efficacy of AI tools through research, says Chris Agnew, director of the Generative AI for Education Hub at Stanford University
This dearth of research exists at a time when AI tool development booms and a strong federal push is on to use AI in their classrooms, Agnew says. Districts are “flying blind, because the best data they have is surveys” or anecdotes from teachers who either strongly support or reject using the tools, he says.
3) Is cognitive thinking encouraged?
It’s important to encourage cognitive thinking when students use AI tutoring. A recent MIT study found potential long-term learning and cognitive skills deficits among those relying on generative AI tools. Relying on AI tools “may unintentionally hinder deep cognitive processing, retention, and authentic engagement with written material,” according to the study.
AI tools perform best when paired with high-quality instructional materials, says DiCerbo. Using human-curated content for AI tutors helps give “better hints and clues for scaffolding and better feedback, and it’s more aligned to what teachers are wanting to teach in the school,” she says.
Students also need help from teachers to ask good questions of AI-powered tutoring tools, DiCerbo says.
In analyzing student transcripts from Khanmigo, Khan Academy “found that there’s a lot of cases where students are just typing ‘idk’ [I don’t know],” or they’re just not asking good questions, DiCerbo says. “There’s some work to be done, I think, to help students get the most out of these tools.”
DiCerbo recommends that school district leaders pilot an AI tool for a certain subject or grade level before scaling it districtwide.
Districts also need to adopt their own acceptable AI use policies. Teachers and students need more clarity on how AI is and isn’t allowed to be used in school settings. Districts should work with communities to develop policies.
K-12 Dive
