As artificial intelligence becomes more common in healthcare, its potential role in mental health services is creating both excitement and concern. For students interested in technology, psychology, or ethics, understanding these issues is increasingly important. AI tools offer possibilities like quicker assessments and customized treatment plans, but they also introduce serious questions about privacy, bias, and the loss of human connection in therapy. According to Stanford, these challenges could influence not only how mental health care is delivered but also how people view therapy itself.
A new Stanford study shows that while some people enjoy interacting with AI systems, therapy includes safety‑critical moments that require human judgment. Senior author Nick Haber emphasizes that current AI models are not equipped to replace trained mental health professionals.
What the Study Examined
To understand how AI compares to human therapists, researchers first reviewed professional therapy guidelines that prioritize empathy, fairness, non‑judgment, and avoiding responses that reinforce harmful thinking. Then they tested five well‑known therapy chatbots—including Pi and Noni from 7cups and the Therapist bot from Character.ai—to answer two major questions:
- Do these chatbots display stigma toward certain mental health conditions?
- How do they respond when users express dangerous or concerning thoughts?
Finding 1: AI Showed Stigma
In the first experiment, chatbots reacted to short stories describing individuals with conditions such as depression, schizophrenia, or alcohol dependence. The systems consistently showed more negative bias toward schizophrenia and alcohol dependence than toward depression. Lead author Jared Moore notes that this bias appears even in newer and larger AI models, suggesting that more data alone won’t fix deep‑rooted stigmas.
Finding 2: AI Mishandled Crisis Signals
The second experiment explored how chatbots responded to signs of suicidal thoughts or delusions. Rather than challenging harmful thinking or steering users toward safety—as a trained therapist would—several chatbots provided information that could worsen the situation. In one example, when a user hinted at suicidal intent by asking about tall bridges after losing a job, the chatbot responded by listing bridge heights, missing the underlying danger. These mistakes are especially concerning because such chatbots have already been used by millions of people.
Implications for the Future
The researchers conclude that AI is not ready to replace human therapists. Therapy is relational and requires emotional sensitivity that AI systems cannot reliably replicate. However, AI could still play a helpful supporting role, such as:
- handling administrative tasks
- serving as a training tool for new therapists
- assisting with low‑risk activities like journaling or coaching
Haber emphasizes that the goal is not to reject AI in mental health care but to clearly define where it can be used safely and where human expertise is essential.
Final thoughts
As you think about the role AI plays in your life, consider:
- Are you using it responsibly?
- Are you using it sustainably?
- Are you avoiding it because you’re unsure of its impact?
- What steps can you take now to build the AI skills you’ll need in the future workplace?
No matter your major or career path, AI will influence the environments you enter. Understanding both its ethical challenges and its practical possibilities is an essential part of preparing for what comes next.
Related Resource: Career Pathways Credentials and Courses on Career Hub
Photo Credit: Unsplash
Source: Stanford Human-Centered Artificial Intelligence. (2024). Exploring the dangers of AI in mental health care. Stanford University. https://hai.stanford.edu/news/exploring-the-dangers-of-ai-in-mental-health-care
