# A randomized trial compared a purpose-built AI tutor with an active-learning class.

A 2025 randomized controlled trial reported that college students using a custom AI tutor learned more in less time than students in the comparison active-learning class.

Canonical: https://brightaifuture.com/signals/signal-ai-tutoring-rct-2025
Format: source-report
Source publication: 2025-06-03
Bright publication: Not established
Substantive update: None recorded
Evidence and review: in-review; source report. Research candidate. The available Stanford repository records month precision only; verify the journal’s exact online publication date and study context before promotion.

## AI’s role

A custom AI tutor delivered course content and responses.

## What is documented

Randomized trial reported more learning in less time than its active-learning comparison.

## Limitations

Tests a particular tutor and course design.

Does not establish effects of general-purpose chat systems.

## Why follow this?

The study offers a bounded test of a particular tutor and course design. It does not show that general-purpose chat systems improve learning or that the result transfers to other students and subjects.

## Provenance and history

{
  "dates": {
    "eventDate": null,
    "publicationDate": "2025-06-03",
    "captureDate": "2026-09-06",
    "lastReviewedDate": null
  },
  "provenance": {
    "origin": "editorial",
    "externalId": "10.1038/s41598-025-97652-6"
  },
  "revisions": [],
  "corrections": []
}

## Original sources

- [AI tutoring outperforms in-class active learning](https://www.nature.com/articles/s41598-025-97652-6)

## Continue exploring

- [How can AI expand learning while preserving human judgment?](https://brightaifuture.com/threads/learning)