Start free trial

The trick: Lab Not Field

Khan Academy sells Khanmigo as an always-available tutor.

A two-year trial found students rarely asked it anything.

Issue 366 October 202612 receipts3 min

Khan Academy markets Khanmigo as an always-available tutor that never gives you the answer.

Before you read on. Your call?

A two-year randomized trial in 18 Tennessee middle schools, by Oreopoulos and Low, found assignment raised maths by 1.3 national percentile ranks per term, about 0.06 to 0.08 standard deviations a year.

The twist

The authors say these gains resemble Khan Academy practice without AI. 96 percent tried Khanmigo, but the median student messaged it in only 17 percent of sessions with a mistake.

18Tennessee middle schools in the two-year randomized trial
1.3national percentile ranks gained per term from assignment
96%of students tried Khanmigo at least once
17%of mistake-containing exercise sessions where the median student messaged it

There’s more to this story.

Membership opens the full investigation, the strongest counterargument and what to do with what you’ve learned.

Start your 30-day free trial →

First membership: 30 days free, then A$89 a year. One introductory trial per customer. Card required; renews annually until cancelled. Cancel before the trial ends to avoid the first charge. Already a member? Sign in

The trick has a name

We call it Lab Not Field: it works in the test conditions, not the deployed ones. You'll see it again. Learn to spot it →

Say this in tomorrow's meeting“In a two-year trial, Khanmigo nudged maths scores up about as much as Khan Academy practice without AI, because students rarely asked it for help.”

Receipts

  1. Context edworkingpapers.com: Assignment raises math achievement by 1.3 national percentile ranks per term, or about 0.06 to 0.08 standard deviations over a school year
  2. Refutes edworkingpapers.com: These gains resemble those from Khan Academy practice without AI assistance.
  3. Context edworkingpapers.com: 96 percent of students tried Khanmigo at least once
  4. Refutes edworkingpapers.com: and in only 17 percent of the exercise sessions in which they made a mistake.
  5. Refutes edworkingpapers.com: Messages that students did send were mostly bare answers or clicks on suggested prompts.
  6. Context edworkingpapers.com: The binding constraint appears to be engagement
  7. Context edworkingpapers.com: a two-year cluster randomized trial in 18 Tennessee middle schools
  8. Supports edworkingpapers.com: the implied effect of a full year of active participation reaches 0.14 standard deviations.
  9. Context edworkingpapers.com: Generative AI has been promoted as the technology that could transform education by providing every student a personal tutor.
  10. Supports khanmigo.ai: Khanmigo is your always-available tutor on just about every topic.
  11. Supports khanmigo.ai: Unlike ChatGPT, Khanmigo never gives you the answer.
  12. Context news.ycombinator.com: AI tutoring with Khanmigo in a two-year school experiment

Open the Receipts Pack → What each source proves, every figure traced, and what would change our verdict.

This story is a stable, citable object. If you can falsify a verdict,tell us. Corrections are loud here.