How many people use AI for mental health questions?
A KFF poll of 1,343 US adults, fielded from February 24 to March 2, 20261, found that 16% had used AI tools for information or advice about their mental health in the past year. Use was highest among younger adults: 28% of those aged 18 to 29 and 18% of those aged 30 to 49. Uninsured adults were about twice as likely as insured adults to have done it (30% against 14%).
The same poll asked what people used it for. About one in ten US adults used AI to understand and compare treatment options, and 7% used it to help decide whether to seek professional mental health care. Among people who had used AI for their mental health, 42% said they later followed up with a mental health professional.
For a good share of people, then, a conversation with an assistant comes before a conversation with a clinician.
Is AI use for health questions growing?
Yes. In a later KFF poll of 2,480 US adults, fielded from May 7 to 31, 20262, 29% said they use AI tools or chatbots for health information and advice at least monthly. KFF reports that this share has nearly doubled in two years, from about 17% in June 2024. That figure covers health questions in general rather than mental health alone, and it shows which way things are moving.
The shift is sharpest among young people. A nationally representative survey of 1,009 Americans aged 12 to 21, published in JAMA Pediatrics in August 20263, found that 19.2% had used AI chatbots for mental health advice by November 2025, up from about one in eight in the same research team’s 2024 survey, which used slightly different wording. Of those who had, 91.7% rated the advice somewhat or very helpful. These are tomorrow’s adult clients, and many already start with an assistant.
Search engines are adjusting too. In February 2026 Bing began showing site owners when their pages are cited as sources in Copilot answers, and its announcement4 put the change simply: visibility “is not only about blue links.”
Where do AI assistants get their information about a practice?
Mostly from the public web. Many assistants, including ChatGPT, Gemini, Copilot and Perplexity, can search the web, read a handful of pages and summarize them, often with links to their sources. Google says its AI Overviews and AI Mode5 may run several related searches to build one answer, and that they show links to supporting pages. OpenAI’s crawler documentation6 describes a search crawler, OAI-SearchBot, that finds the websites ChatGPT shows in its search answers.
The material available to them is whatever describes your practice in public: your website, your Google Business Profile, your Psychology Today listing and any other directory profile. Google’s AI optimization guide7 says a Business Profile can help a business appear in AI responses as well as in ordinary search results.
Two practical points follow. An assistant can only pass on what it can find and read. And when your website and a profile disagree about your fee or whether you are taking new clients, it has no reliable way to tell which one is current.
What can a practice do to be described accurately?
- Give each service its own page. Say who it is for, whether you see people in person, online or both, where you are, which states you can see clients in, and how to start. A page that answers “do you see couples in Pasadena?” in plain words is easier for a person and a search tool to use than a long list of specialties.
- State fees and payment plainly. Your session fee or range, whether you hold sliding-scale spots, whether you provide superbills, and which insurance you take, if any.
- Make your profiles match your website. Practice name, address, phone, telehealth area, fees and availability should read the same on your site, your Google Business Profile, Psychology Today and every other directory. Google’s AI features guidance lists keeping Business Profile information up to date among its recommendations. When something changes, change it everywhere in the same week.
- Add structured data that matches the page. Structured data is a short block of code that labels facts such as your practice name, address and services. Google describes it8 as a way to give its systems explicit clues about what a page means, and asks that it match the visible text.
- Keep pages readable by people and software. Real text rather than text inside images, clear headings, links that say where they go, and a site that works on a phone and with a screen reader. Check that your site does not block the crawlers you want. OpenAI says sites that block OAI-SearchBot will not be shown in ChatGPT search answers. Training is handled by a separate crawler, GPTBot, so you can decline training and still allow search.
- Check what assistants say. Ask ChatGPT, Gemini, Copilot and Perplexity what a prospective client would ask: your practice name, and “therapist for [what you treat] in [your city].” Ask each more than once, note the date, and write down anything wrong. Google notes9 that results can vary with time, location and personalization, so one answer is a sample. When an answer is wrong, find the page it relied on and correct that source.
Can you control what ChatGPT says about your practice?
No one controls what an AI assistant says, so good work focuses on the parts you can control. Google states that it may still not crawl, index or show a page that meets all of its requirements, and assistants word their answers a little differently each time.
The material they work from is yours to shape. Accurate pages, matching profiles, a site that tools can read, and regular checks give every assistant the best available version of your practice. The same work helps anyone who reads your website directly or finds you through Google, a directory or a colleague’s referral. Our SEO for therapists guide puts that work in order over 90 days.
If you would like help with that work, our Practice Visibility page explains how we approach it. If you are comparing providers, our guide to what SEO services for therapists cost lists published prices and a contract checklist.