Current solutions that are being developed to reduce the risks of using AI for mental health support include:
- Regulations and Laws
- Educating the public about using AI in this way
- Human Oversight and Detection Systems
Regulations and Laws
The first solution is to have regulations and laws to limit what chatbots can do. For example, recent legislation introduced by lawmakers prevents AI chatbots from impersonating licensed professionals such as doctors and therapists. This is known as the CHATBOT Act (H.R. 7985). As stated in the article, the goal of this act is to stop AI systems from “impersonating licensed professionals in the medical, legal, and financial fields,” which helps prevent users from being misled into thinking they are receiving legitimate medical advice [12]. These laws and regulations are important because it can stop the AI from giving false information to users. And can reduce the amount of trust the user has in the chatbot.
Educating the public
The second solution is educating the public about AI. The public must have a good understanding of AI and AI literacy. If people are taught about the risks of AI and how the system actually works, then they will be less likely to treat it as a therapist. When people are aware of the risks, they are less likely to rely on AI for mental health issues. Then they will be more likely to go to trained professionals instead, who can give them the support they need. LinkedIn and other websites offer AI literacy courses that teach users how AI systems work and the risks of relying on them too much.
Knowing the current efforts to solve the issue, what are some other things we can do?
Human Oversight and Detection Systems
Another current solution being developed is stronger human oversight and detection systems for AI chatbots. Companies are working on ways for chatbots to recognize signs of serious mental health problems, such as depression, self-harm, or suicidal thoughts. If these happen, the chatbot will encourage the user to contact a licensed mental health professional instead of continuing the conversation. OpenAI claims that ChatGPT is being updated to better recognize emotional distress and is directing them to professional resources [13]. This solution is important because AI cannot fully understand human emotions or safely handle these situations. Having human professionals involved can reduce the risk of AI giving harmful advice and make sure users receive proper care when they need help.