Artificial intelligence, AI, large language models, chatbots, ChatGPT, Claude, Gemini… Unless you’ve been offline for the past five or so years, you know what I’m talking about. No matter how you feel about AI, it’s increasingly becoming part of everyday life. As I’ve had more and more conversations about AI in my daily life, it occurred to me that it might be useful to share how I currently see it fitting within The Seven Challenges framework. (Note: For simplicity, I’ll refer to AI chatbots such as ChatGPT, Gemini, Claude, etc. generically as AI.)

A study conducted by the RAND Corporation in 2025 and published a few months ago found that nearly 20% of people ages 12 to 21 had used an AI chatbot for mental health advice. That’s up from 13% in 2024 and is nearly the same rate as 12- to 21-year-olds receiving counseling from a mental health professional. About 6% of adolescents reported using chatbots daily or almost every day. The Pew Research Center also recently reported on adult AI chatbot use, finding that 10% of adults use chatbots for emotional support or advice.

There are several ways to interpret these numbers. It may be that some people who seek support from mental health professionals are also seeking support from AI. It may also be that those who aren’t seeking professional support are increasingly turning to AI chatbots as an alternative. Regardless, the data highlight that a sizable and growing number of adolescents and adults are using chatbots for mental health support. And, I’d be shocked if those numbers decrease anytime soon.

You may already be doing this, but if you aren’t, I think it is time to start talking to all of your clients about AI because there is a good chance it is a part of their life. For new clients, I think this conversation pairs well with learning about what other sources of professional, informal (e.g., friends), or emotional support your client is getting.

The key is to have a relaxed and open conversation and resist the urge to rush in with our perspective. This might feel challenging if you have your own, potentially strong, opinions about AI (positive or negative!). But I’m sure you can do it. Much like you would approach a conversation about someone’s drug use or friendships using a Mastery Counseling approach, you want to bring the same open, relaxed, accepting, and nonjudgmental perspective.

That means not jumping in with, “You know, AI is really dangerous. You need to stop using it. It’s not trained to help.”

There may be legitimate concerns behind those statements, but leading with a negative perspective would likely cause a client to shut down. Instead, start with an open and honest conversation (Challenge 1).

If your clients are using AI, get curious. Much like drugs or support from friends, it is vital to understand and validate the needs that your clients are fulfilling with AI (Challenge 2). They may be using it for information, reassurance, emotional support, connection, or simply because it is free, easy, and always available. You might learn they’re using it because it provides unconditional support and empathy. If you dig a little deeper, you may discover that they don’t have any other source for that kind of support in their life. All of a sudden, you have an important clinical issue.

When clients freely discuss what they’re talking with AI about, it’s an opportunity to dig deeper. For example, AI might suggest a client use “a small bit of meth” to get through the week. It’s worth discussing the benefits (Challenge 2), harms, and potential future harms (Challenges 3 and 5). Discussion of benefits would likely reveal underlying motivation for particular behaviors. In the meth example, the motivation could be to help an overtired person get through the work week and not lose their job (Challenge 4). The conversation about AI has given you a new way to understand what’s going on in your client’s life.

The same thing can happen with alcohol and other drug use. Clients might be using an AI chatbot to figure out whether their drinking or drug use is “really a problem.” They might ask how to cut back, how to stop using, what withdrawal might be like, or even how to reduce the risks associated with using a particular substance.

I’d suggest starting with curiosity. What were they hoping to learn? Why did they ask AI instead of someone else? What did they find helpful? Was there anything about the response that concerned them?

This can be particularly useful from a harm reduction perspective. For example, a client who is using AI to figure out how to drink less or use more safely may already be thinking about making a change. We can start with what they’re trying to accomplish. Then, we can ask what they have heard from AI and complement it with our input to support them in making an informed decision (Challenge 6).

Of course, being non-judgmental about AI doesn’t mean we should assume that everything AI states or recommends about particular behaviors or situations is accurate or safe. We know our clients personally and can ask questions for clarification. We are not basing our input on broad algorithmic generalizations. We have important input to offer. Like with their drug use, we want to validate the needs that AI is helping them fulfill (Challenge 2). We also want to discuss the harms or potential harms (Challenge 3).

Our goal is to help clients understand why they’re using AI, what they’re getting from it, supplement with our input, and then assist them in making wise decisions (Challenge 6).

If a client tells you they don’t use AI chatbots, it may also be worth exploring a little further. Are they truly not using them or are they uncomfortable telling you they are? Adolescents in particular may be accustomed to butting heads with adults about the use of AI and could be reluctant to open up about it (just like clients can have a hard time opening up about their drug use or friends). This doesn’t mean you should interrogate them or assume they’re hiding something. Just stay curious.

Finally, you may have noticed that I haven’t said much about my own opinion of AI chatbots. That’s intentional. At this point, the research on many of these tools is limited, and most widely available chatbots, including ChatGPT, Gemini, and Claude, were not developed specifically to provide mental health treatment. From what I can tell, they can sometimes function a lot like a friend who is always on your side. We all like having that person in our corner. But sometimes we also need the friend who helps us take the other person’s perspective, points out when we might be missing something, or tells us something we don’t necessarily want to hear.

When I have used these tools to test how they respond to made-up clinical concerns, they generally point out that they aren’t a replacement for professional or medical care. Some of the suggestions seem legitimately helpful and others do not. So, I imagine some people may find AI chatbots to be genuinely useful, while others may not.

As counselors, let’s stay neutral on AI. It is there, and our clients may be using it. We want to be helpful any way we can. We just need to be willing to ask and have an open, curious, nonjudgmental conversation. We may find the conversation tells us a lot about what is really going on in our clients’ lives.