The more we learn about artificial intelligence (AI) and the effects it is having on mental healthcare, the more we understand the power of AI technology when used correctly. From providing support for both patients and therapists to freeing up time and resources so that behavioral therapists may solve problems with the human touch that machines simply cannot, AI will continue to offer new opportunities for improvements in mental healthcare.

Tools powered by artificial intelligence (AI), such as ChatGPT, are offering new opportunities to people affected by mental health conditions.

Chatbots can respond to prompts conversationally, creating a more engaging and intuitive user experience. Users can message with the chatbot to talk about moods, feelings, and thoughts, and the digital technology can offer real-time advice and guidance for where to seek support, emulating a face-to-face therapy-like experience.

Mental health is a growing and important societal concern, and lack of care for those who need it is a top reason why artificial intelligence is being viewed as an increasingly promising piece of the mental healthcare puzzle. Access to information, resources and advice have been made dramatically easier by AI, but users should exercise caution when using it for mental health information.

Here are 4 things to keep in mind when using AI for mental health:

1. AI can be beneficial for mental health — when used correctly. As mental health concerns continue to rise, ChatGPT can be leveraged to support users by offering support, providing coping strategies and self-care suggestions, and reminding patients of follow-up appointments. It can also be used by providers to give better insights into their work, to ensure high standards of care and to help improve training.
2. It should not be used to replace the human connection. While ChatGPT can generate human-like responses and give detailed information to help you learn about mental health, it cannot diagnose an illness. It cannot empathize and it cannot understand the complexities of human emotions. It can suggest you seek help from a medical professional — but you should not rely on it for sound medical advice. Always defer to a trained professional, whether it be an in-person therapist or a telehealth service.
3. Your “prompt engineering” makes a difference. The more specific your prompts are, the more targeted the responses become. If you give a broad overview for a topic you’re seeking information on, you will receive more generic responses rather than specific information. You can create better prompts by including symptoms, what objective you’re seeking and any pertinent, non-personal details to the topic you’re interested in learning about.
4. You’ll still need to fact-check and keep your sensitive information private. Remember that as with anything health-related, you’re seeking accurate and verifiable information. Ask for sources and for it to cite specific studies — and check those sources. Blindly trusting the information can be very risky as the algorithm is not set up to fact check every piece of information. And on that same note, remember to keep your personal information out of the conversation. When feeding ChatGPT prompts, it can be easy to include your private information by mistake — but keep in mind that all of that information is stored for future use. Instead of including personal identifiers or sensitive health information in your prompts, keep things more broad and save the personal information for chats with your medical professional.

While AI technology like ChatGPT will not take the place of a medical professional, it can be a powerful support tool in mental healthcare, by enhancing efficiency and accessibility to care. Some may even prefer chatting with a bot, feeling less stigma in asking for help from a machine. However, it’s important for those seeking help to remember that AI technology should be an ally in the collaborative care necessary for a successful mental health journey.