Millions of people are already using AI for mental health support, whether they'd call it "therapy" or not. 

Adam Dodge and Dr. Saed D. Hill take an honest, non-judgmental look at how AI is showing up in mental health care, what it does well, where it falls short, and what it means for the future of therapy.

Three Ways AI Shows Up in Mental Health General-purpose chatbots used informally for support; purpose-built therapy apps with clinical input; and AI companions that "shape shift" across roles: girlfriend, therapist, trainer, etc. Spoiler: blurring those lines is not a great idea.

Why AI Therapy Is Everywhere ChatGPT handles roughly 18 billion messages a week. If just 2% involve mental health, that's 342 million therapeutic conversations weekly, potentially making it the largest mental health provider in the country. That's not an accident. It's a direct response to a broken system: nearly 50% of people who need care can't access it.

What AI Does Well Dr. Hill makes the case that AI can genuinely shine at skills-based, manualized approaches like CBT, helping people recognize thought patterns, challenge distorted thinking, and build coping strategies. He tested it himself on Character AI and found it impressive for skill-building. Short-term support for anxiety and depression? Research backs it up.

Where It Can Get Dangerous AI can't read the room. It misses body language, tone, and the unspoken signals human therapist picks up in real time. It can't do repair work. In a crisis, the stakes are life or death, and an active lawsuit against OpenAI involving a minor's death makes that concrete. A Stanford study on LLMs and therapy reached a clear conclusion: not a replacement. And none of this is HIPAA-protected.

The "New Relationship Energy" Trap 24/7 availability, zero judgment, and constant validation feels great at first. But endless affirmation without challenge can stunt emotional growth. Real growth often requires tension, rupture, and repair. Those are things AI doesn't offer.

The Hybrid Future Rather than competition, the hosts see AI as a bridge between sessions: a reflection tool, a way to review skills between appointments, or eventually an AI trained on your own session transcripts to reinforce what you and your therapist are working on.

If you're considering an AI therapy app:

• Was it purpose-built for mental health?
• Who built it? Clinical expertise/credentialed advisory board matter.
• Was it researched and tested?
• What data are they collecting? How is it protected? (No HIPAA = no confidentiality guarantee.)

If you're a therapist:

• Ask clients about their AI use
• Learn the technology

Research Referenced

OpenAI / ChatGPT message volume statistics

Mental health access gap: ~50% of people who need care can't reach it

1 mental health provider per 10,000 people seeking care; 1,600 patients per available provider

"Therapy deserts" (geographic mental health access gaps)

Stanford study: LLMs and therapeutic replacement

Lawsuit against OpenAI involving a minor's death (Raine v. OpenAI)

Mailbag
We are putting together mailbag episodes and want your questions. If something from this episode — or any episode — sparked a question you want us to dig into, send it our way. Nothing is off limits.

Subscribe & Review
If Why Humans? is a podcast you find yourself thinking about after the episode ends, the best thing you can do is subscribe and leave a review wherever you listen. It helps more people find the conversation.

Follow Us
Stay in the loop between episodes. We share clips, resources, and things that make us think. 

Instagram — @thewhyhumanspodcast