Provider: Amira
Program Description
This program pairs an AI-based reading tutoring platform with a live human helper. Importantly, the human helper does not teach reading directly. Instead, they help keep students engaged and motivated, assist with any technical issues, and manage the group, while the AI platform delivers the actual reading lessons and feedback. Each helper worked with a group of two to five students. In one district, helpers worked with students after school; in the other, older students at the same school served as helpers during the school day.
Findings
- Significant positive effect on engagement: Students working with a human helper read almost twice as many stories per week, but overall engagement remained low. Reading test scores, a secondary check, showed no meaningful improvement in student learning.
- Students with a human helper used the platform slightly more than the comparison group: about one extra minute per week in District A, about four extra minutes per week in District B. This translated to average weekly usage of 2.18 minutes in District A and 5.23 minutes in District B.
- Without a human helper, students who used the platform on their own tended to already be stronger readers and less likely to have a disability.
- No cost data was reported. Comparison-group students used the AI platform entirely on their own, with the same scheduled access as the group that got a human helper, but no adult support.
- Low overall usage limits interpretations of the impact of a human helper. Small, uneven samples across after-school vs. in-school sites make those comparisons unreliable.
| Research Partner | NSSA |
| Study Year | 2024-25 |
| Study Location | Two school districts |
| Modality | AI Platform with In-Person Human Support (the reading instruction is delivered via an AI program, while a helper provides in-person encouragement and support |
| Study Design | RCT (Randomized Controlled Trial) – two separate versions of this study ran, one in each district |
| Total Sample | 355 students total (174 in District A, 181 in District B) |
| Characteristics of Treatment Group | District A (n=174): Female 52% Economically Disadvantaged 87% Students with Disabilities 11% English Language Learner 21% Black 26% White 35% Hispanic 58% Multiple/Other 39% District B (n=176–181): Female 54% Economically Disadvantaged 91% Students with Disabilities 26% English Language Learner 31% African American 67% Asian 11% Hispanic 14% Multiple 3% White 3% |
| Subject | Reading/ELA |
| Grades | 1, 2, 3, 4, 5 |
| Average Actual Dosage | 0.3 hours (varied by district) |
| Scheduled Dosage | 12.8 hours (varied by district) |
