Khan Academy rolls out AI tools

Khan Academy’s AI push is getting more concrete. Not just “we have a chatbot now” concrete — more like better measured tutoring on the K–12 side, and an actual degree plan on the higher-ed side. The gap it’s trying to fill is obvious: school support is uneven, college is expensive, and most AI education talk still lives at the demo stage. In the last few weeks, Khan Academy has moved both pieces forward. (blog.khanacademy.org) ### What changed this week? The near-term update is Khanmigo itself. On May 1, Khan Academy said its latest round of product work improved a key learning metric — “next-item correctness,” meaning whether a student gets the next problem right without help — by 6 percentage points after six months of testing from (blog.khanacademy.org)tries to measure whether the student actually learned something. (blog.khanacademy.org) ### Why is that a bigger deal than new badges? Because the interesting part is not the surface layer. A lot of coverage around AI study tools drifts toward progress bars, gamification, and instant feedback. But Khan Academy’s own write-up puts the emphasis elsewhere — response speed, whether the tutor gives answ(blog.khanacademy.org), it is trying to tune the tutor so it behaves more like a real instructor and less like an answer machine. (blog.khanacademy.org) ### How is Khanmigo supposed to teach? The core idea has not changed. Khanmigo is meant to guide students with questions and hints rather than just dumping solutions. Khan Academy says it tracks whether the exchange stays cognitively active and whether students can transfer that help to the next problem on their(blog.khanacademy.org)a student smarter. Khan Academy is trying to prove the second thing. (blog.khanacademy.org) ### So where does the college angle come in? That is the Khan TED Institute. On April 14, Khan Academy, TED, and ETS announced a joint higher-ed venture aimed at the AI era. The plan is a mostly competency-based program with an intended price under $10,000, and applications are expected to open in 12 to 18 month(blog.khanacademy.org)igh cost tied to a time-based model that often says little about what a graduate can actually do. (ets.org) ### What would students actually study? The first program is organized around three buckets: academic foundations, applied AI work, and communication and leadership. The applied side includes things like AI-assisted app development, financial modeling, AI agents, and team deployment projects. The partners(ets.org)h is pretty clear — teach practical AI skills, then make the proof of skill legible to employers. (ets.org) ### What is the catch? The K–12 and higher-ed pieces are at very different stages. Khanmigo is live and being iterated with measured outcomes. The degree idea is still a plan. Under-$10,000 sounds disruptive, but a cheap credential only works if employers trust it and students believe it opens doors. That part is not solved by good curriculum alone. (blog.khanacademy.org) ### Why does this matter now? Because Khan Academy is crossing a line from “AI helper” to “AI-shaped education system.” One piece tries to improve how students learn day to day. The other tries to change what counts as proof that they learned it. If both work, Khan Academy will have done more than roll out tools — it will have built a pipeline from tutoring to credentialing. (blog.khanacademy.org) ### Bottom line? The real story is not flashy classroom AI. It is that Khan Academy is trying to make AI legible in education — first as a tutor you can measure, then as a pathway you can price, structure, and maybe eventually trust. (blog.khanacademy.org)