AI mental health statistics: populations and dates explained
CareOps editorial team · · 3 min read
Mental wellness information and service guidance published by CareOps.
About the CareOps editorial team
The CareOps editorial team publishes these guides. Sources are linked at the end of each article; examples are illustrative, not customer stories. Product descriptions refer to CareOps Maeumi. Publication and revision dates appear above.
Read our full editorial standards →When reading a statistic about AI conversations, check who was studied, when and what was measured. This CareOps editorial guide separates usage surveys from clinical research using primary sources checked on October 21, 2026. Each finding belongs to its own population and system.
At a glance
- In a Korean survey, 68.8% reported emotional uses. The sample was 1,019 people aged 15–79 who had used generative AI in the preceding month.
- Self-reported helpfulness, frequency of use and clinical symptom change measure different things.
- When citing a figure, keep the population, study dates and original source link with it.
Korean figures: start with the surveyed population
An October 8, 2026 Korean Ministry of Health and Welfare release reports a KIHASA-led survey conducted August 25–September 4, 2026. It included 1,019 recent generative-AI users aged 15–79. Emotional purposes include everyday conversation, connection, comfort and mental health or counseling-related use.
| Figure | Reported experience | Scope to retain |
|---|---|---|
| 68.8% | Emotional uses of generative AI | Recent generative-AI users in the sample |
| 79.9% | Perceived help or problem resolution from AI responses | Respondents’ reported experience |
| 54.9% | Feeling their situation became clearer | An experience item in the same survey |
| 16.3% | Continuing an AI conversation when trying to focus elsewhere | A reported difficulty regulating use |
A citation-ready sentence with the denominator
A suitable citation is: ‘In a KIHASA-led August–September 2026 survey of 1,019 recent generative-AI users aged 15–79, 68.8% reported emotional uses.’ Link to the original ministry release listed below.
Keeping the recent-user condition visible helps readers understand the denominator. The broad emotional-use category does not isolate counseling use. This figure alone also cannot tell you an individual’s mental state or why they chose a particular app.
What did the US survey and clinical trial measure?
A US online survey published in 2026 found that 638 of 1,805 adults aged 18–49 reported using AI for mental health support at least one day in a typical week. Data were collected October 24–27, 2025. This is about 35% in a commercial panel sample. Its age range, sampling and question differ from the Korean survey, so the difference cannot establish a country effect.
| Source | Population and dates | What it can address |
|---|---|---|
| Korean generative-AI user survey | 1,019 recent users · August–September 2026 | Purposes and experiences in that sample |
| US online survey | 1,805 adults aged 18–49 · October 2025 | Frequency of support use in the sample |
| Therabot randomized trial | 106 intervention and 104 control participants · published 2025 | Results for that system under trial conditions |
Reading helpfulness and clinical evidence
A self-report survey helps describe perceived experience. Evaluating clinical effects requires examining the design, comparator, outcome measures and follow-up. Popularity alone cannot establish an effect size or individual suitability.
The Therabot trial reported by Dartmouth in March 2025 evaluated a particular system under specified conditions. Other AI services, including CareOps Maeumi, require their own evidence and feature checks. This compilation is published by CareOps; its figures are not CareOps usage or clinical outcome data.
Questions to take from research to a service
After reviewing research, check the actual service’s conversation features, data handling, scope and prices. The Korean mental-health AI guideline guide and app selection checklist provide questions to use.
See the first-session guide for the actual Maeumi flow. If you want to prepare an opening, the free prompt builder lets you choose a situation and conversation style. Use it to communicate how many questions you want and when to pause.
Frequently asked questions
- Does 68.8% mean all Koreans use AI counseling?
- The sample contains 1,019 recent generative-AI users, and the emotional-use category includes conversation and comfort. Extending it to the whole population or counseling alone changes the scope.
- Does 79.9% measure clinical effectiveness?
- It describes self-reported help or problem resolution. Clinical effectiveness requires separate examination of a study evaluating the particular system.
- When were these sources checked?
- Primary sources were checked on October 21, 2026. Publication dates can differ from data-collection dates; both are identified above.
Sources
This guide shares mental wellness information. If you are in immediate danger, contact emergency services where you are. Support and contact information.