The Friend Who Never Pushes Back: What We're Noticing About AI and Mental Health

Something new has been showing up in my conversations with patients: a third party. Not a spouse, not a friend, not a therapist. A chatbot.

I'm Christopher Schuman, a board-certified psychiatric nurse practitioner serving patients in Texas and Washington. This month I want to talk about AI and mental health, starting with what I'm actually seeing and one thing everyone using these tools should understand.

How People Are Actually Using AI

Most of what I encounter is surface level and fairly practical: people asking about medications and side effects, uploading lab work and asking what it means, using AI at work. Reasonable uses, with caveats we'll get to later this month.

But there's a subset of people using AI as something more like a third person in their lives. Not for companionship exactly, but for judgment and reasoning. What do you think about this situation? Was my tone in this email appropriate? Here's what they texted back, what do you think they meant? The chatbot becomes a kind of always-available second opinion on being human.

That's where things get complicated.

The Bias Built Into the Machine

Here's the thing I most want people to understand: these tools carry a strong degree of confirmation bias. They're designed to sound encouraging and engaging, because that style of language keeps conversations going. As I've put it to patients, an AI chatbot is a friend who doesn't want to lose your business.

Ask a leading question and you'll usually get the answer you were hoping for. "That was a good idea, right?" tends to produce agreement, even when, in retrospect, it was a terrible idea. Unless someone knows to deliberately prompt for neutrality, asking how an objective third party would see the situation, the machine's default is to validate.

Now consider who's most likely to lean on that. Often it's people who are conflict avoidant or anxious at baseline, seeking reassurance. The tool hands them exactly the reassurance they came for, every time, at 2 AM, without judgment. That's not wisdom. That's a mirror with a kind voice.

What the Machine Doesn't Know

There's a second limitation that matters just as much: an AI only responds to what it knows, and it doesn't know you.

It doesn't have the context of your history, your patterns, or the other person's personality when it interprets a text message for you. If you've vented about your relationship a dozen times, it takes that record as the whole story and builds its advice on it. It can even confidently produce information that's simply wrong, invented facts that sound authoritative until you try to verify them. Anyone who's asked a chatbot for a store's hours and driven to a closed building has met the small version of this problem. The big version shows up when the question is whether to end a relationship or cut off a family member.

The Point of This Series

None of this makes AI evil, and this month isn't a scare campaign. These tools can be genuinely useful, and later in the series I'll talk about where they shine and how to use them well. But a tool this persuasive, this available, and this agreeable deserves clear eyes, especially from anyone using it in a vulnerable moment.

The starting principle is simple: an AI can process your words, but it cannot know your life. Keep that distinction, and the tool stays a tool.

The Bottom Line

If you've been asking a chatbot the kinds of questions you'd once have taken to a trusted friend or mentor, you're not strange, and you're certainly not alone. Just know what's on the other end: a system built to agree with you, working from only what you've typed. Some questions deserve more than that.

Want a Second Opinion That Actually Knows You?

New patient consultations available within 2-4 weeks.

Insurance: We accept most major insurance plans and offer transparent pricing for self-pay patients.

Crisis support: If you're experiencing a mental health emergency, please call 988 (Suicide & Crisis Lifeline) or go to your nearest emergency room.

Christopher A. Schuman, MSN, ARNP, PMHNP-BC, is a board-certified psychiatric nurse practitioner and founder of Veritas Behavioral Health, serving patients in Texas and Washington.