How AI is changing the mental health space

• Key Insight: See why 49% of workers with mental health conditions are using ChatGPT for support.
• What's at Stake: Struggling employees risk worsening mental health conditions if employers provide the wrong digital tools.
• Forward Look: Expect AI technology to improve while still avoiding employee diagnosis and treatment.
  Source: Bullets generated by AI with editorial review.

The mental health space is no stranger to emerging technology, with many platforms turning to AI to expand and enhance their capabilities. But what could the growing role of AI mean for the future of mental health care overall?

According to a recent survey by the nonprofit Sentio Marriage and Family Therapy program, 49% of people who use large language models like ChatGPT and report having an ongoing mental health condition said they use AI for mental health support. Among them, 73% said they use it to manage anxiety, 60% for depression support, 58% for emotional insight, 56% to improve their mood, and 35% to feel less lonely. 

As leaders look for new ways to leverage technology to support their workforces, they'll need to carefully vet and implement AI tools to ensure they deliver positive outcomes.

"The reason AI has become very popular in mental health is because it's significantly more convenient for people," said Dr. Jon Cohen, CEO of virtual mental health platform Talkspace. "A virtual mental health visit is easier than scheduling an appointment, leaving the house, and sitting in an office for an hour." 

Read more: Leading benefits professionals buying into AI, with caution

Among the people who are using AI for mental health support, 63% said the technology has improved their mental health, according to Sentio, with 87% saying they found the practical advice helpful or very helpful. Accessibility is the most commonly cited reason for using AI at 90%, the data found, followed by affordability at 70%. And while 39% say the chatbots are about as helpful as human therapy, 36% even say they find the technology more helpful than human therapists, proving technology's usefulness in progressing organizations' commitment to mental health efforts. 

However, the problem isn't whether the technology is effective, according to Cohen, it's that most of the commonly used AI platforms weren't designed to support people through mental health crises and their resources are often limited. As a result, there have been several cases where a chatbot played a part in a user's worsening condition. 

"People are clearly comfortable using these agents," Cohen said. "For employers who are trying to help their employees, our plea is: OK, if they're going to use it. Use something that's safe." 

Choosing the right AI

To ensure that the AI resources they're introducing are as safe as they are successful, Cohen advised leaders to only work with platforms and organizations within the mental health space or that have significant clinical oversight. For example, Talkspace has trained its AI Tee using years of mental health data, including millions of diagnoses, psychiatry interactions, treatment plans and outcomes. The platform also has HIPAA protections and proprietary algorithms designed to flag potential risks such as suicide, self-harm, violence, substance use and other mental health conditions. Those alerts are then monitored by clinical staff who can intervene when necessary, including by connecting users with a therapist or direct them to emergency care. 

Read more: How AI can help care management programs to scale

"I've always said that the silver lining with AI is that it has democratized the access to mental health support," Cohen said. "Millions of people now, who obviously want help, are seeking help. The issue is how do we provide it responsibly."

According to Cohen, AI can continue to positively expand the reach of mental health efforts; it will just take a little more work from benefit and HR leaders to invest in the right kind of tech and the right kind of help. 

"Like most things, technology will continue to improve," Cohen said. "It still won't diagnose, or advise, or treat employees — we'll stay away from that. But it really can help people get better."