# Study finds lower independent performance after AI assistance is removed

> Three online randomized experiments involved 1,222 people; the findings do not establish effects beyond the short tasks tested.

- Canonical URL: https://newsverum.com/en/news/ai-assistance-reduces-persistence-study-october-2026
- Publisher: Newsverum
- Section: Technology
- Country: United States
- Published: 2026-10-11T15:19:58+00:00
- Updated: 2026-10-12T06:19:19+03:00
- Event / data period: 2026-10-09
- Other editions: [Türkçe](https://newsverum.com/tr/news/ai-assistance-reduces-persistence-study-october-2026.md), [العربية](https://newsverum.com/ar/news/ai-assistance-reduces-persistence-study-october-2026.md), [Français](https://newsverum.com/fr/news/ai-assistance-reduces-persistence-study-october-2026.md), [Español](https://newsverum.com/es/news/ai-assistance-reduces-persistence-study-october-2026.md), [Deutsch](https://newsverum.com/de/news/ai-assistance-reduces-persistence-study-october-2026.md), [Português](https://newsverum.com/pt/news/ai-assistance-reduces-persistence-study-october-2026.md)

A team that included UC Berkeley researchers presented an updated, peer-reviewed study at the Conference on Language Modeling during the week of October 9. Three online randomized experiments enrolled 1,222 participants: two used fraction problems and a third used SAT reading-comprehension questions. AI access was later removed from the assisted groups.

After removal, participants who had AI access solved fewer test questions correctly than control groups; skipping also rose in some experiments. The findings are limited to short, structured tasks. The study does not establish that everyday use causes long-term or permanent cognitive harm; more research is needed to measure how different forms of assistance affect learning.

## Data

### Experiment participants

Participants recruited through an online research platform in the United States for three randomized experiments; final analysis samples were smaller.

Methodology: Participants were recruited through the US-based Prolific online research platform. Values are total recruited sample sizes by experiment; analyzed groups were smaller after attention and ability checks.

| label | value | unit | period | basis |
|---|---|---|---|---|
| Participants in fraction-solving experiment 1 | 354 | people | 2026-10-09 | observed |
| Participants in fraction-solving replication | 667 | people | 2026-10-09 | observed |
| Participants in SAT reading-comprehension experiment | 201 | people | 2026-10-09 | observed |

Sources: <https://arxiv.org/abs/2604.04721>, <https://news.berkeley.edu/2026/10/09/using-ai-for-just-10-minutes-erodes-your-ability-to-persist-at-hard-things/> · Download data: [JSON](https://newsverum.com/api/v1/articles/ai-assistance-reduces-persistence-study-october-2026/data/experiment-cohort-sizes?lang=en&format=json) · [CSV](https://newsverum.com/api/v1/articles/ai-assistance-reduces-persistence-study-october-2026/data/experiment-cohort-sizes?lang=en&format=csv)

## Sources

- [UC Berkeley News](<https://news.berkeley.edu/2026/10/09/using-ai-for-just-10-minutes-erodes-your-ability-to-persist-at-hard-things/>) (Source checked: 2026-10-11; Published: 2026-10-09)
- [arXiv paper, version 5 revised October 3, 2026](<https://arxiv.org/abs/2604.04721>) (Source checked: 2026-10-11; Published: 2026-04-06)

## Correction history

- 2026-10-12T06:19:19+03:00: Wording in this article's language editions was corrected to retain the article's full information and scope.

---

Cite the canonical URL above. [Editorial standards and corrections](https://newsverum.com/en/editorial)