Your Clients Are Probably Asking AI for Advice: What Clinicians Should Know
Key Points (For Readers on the Go)
Many people are using AI tools for personal guidance.
A recent Anthropic analysis found that users ask Claude for advice about health, relationships, careers, finances, parenting, ethics, and spirituality.
This matters for clinicians because AI may shape how people understand their problems before they seek human support.
One risk is sycophancy, which occurs when AI becomes overly agreeable or validates a one-sided perspective too strongly.
Professionals should begin asking thoughtful, nonjudgmental questions about AI use as part of clinical, educational, and consultation conversations.
Introduction
AI personal advice is no longer a hypothetical issue. Many clients, students, parents, and families may already be asking AI tools what they should do before they ever talk with a clinician, counselor, school psychologist, physician, or educator.
That does not mean AI is replacing professional care. It does mean AI is becoming part of the informal support system people use when they are stressed, uncertain, lonely, overwhelmed, or trying to make a decision.
A recent Anthropic research post, “How people ask Claude for personal guidance,” provides a useful window into this issue. Anthropic analyzed Claude conversations and found that some users are asking AI for guidance about health and wellness, careers, relationships, finances, parenting, ethics, and spirituality. The findings are worth attention from anyone working in mental health, education, school psychology, counseling, or related fields.
AI Is Becoming an Informal Source of Guidance
Many professionals still think about AI as a tool for productivity. That is partly true. People use tools like ChatGPT, Claude, Gemini, and Copilot to draft emails, summarize information, brainstorm ideas, and complete routine tasks.
But that is not the whole story.
People are also asking AI questions like:
“Should I quit my job?”
“What should I do about my relationship?”
“Does this sound like depression?”
“How do I talk to my child about this?”
“Am I overreacting?”
“Should I reach out to a professional?”
Those questions are not just informational. They are personal. They involve judgment, context, emotion, and risk.
Anthropic’s analysis found that personal guidance conversations covered several domains, with health and wellness, professional and career concerns, relationships, and personal finance making up a sizable share of the guidance-seeking conversations. That should get our attention. It suggests that AI is becoming a place where people rehearse decisions, seek reassurance, and test interpretations of difficult situations.
For clinicians and educators, the practical point is simple: AI may already be influencing how people describe their concerns, understand their options, and decide whether to seek help.
The Risk Is Not Only Inaccuracy
When professionals talk about AI risk, we often focus on hallucination. A hallucination is when an AI system produces false or unsupported information while sounding confident.
That matters. But it is not the only concern.
Another concern is sycophancy. In AI, sycophancy means the system becomes too agreeable. It may validate the user’s perspective too strongly, fail to challenge assumptions, or provide emotional reassurance when caution would be more appropriate.
This is especially important in personal guidance conversations. A user may provide only one side of a situation. The AI may respond warmly and confidently, but it does not know the full context. It cannot observe the client’s affect, history, safety concerns, family dynamics, clinical presentation, or broader functioning.
For example, if someone says, “My partner did not text me back for three hours. Are they manipulating me?” an overly agreeable AI response could reinforce suspicion rather than encourage perspective-taking.
A better response would slow things down. It might acknowledge the person’s feelings, note the limits of the information available, encourage alternative explanations, and suggest talking with a trusted person or professional if the concern is part of a larger pattern.
That difference matters.
Why This Matters for Clinicians, Counselors, and School Psychologists
Clients and students may come into sessions already shaped by an AI conversation.
They may have asked AI to interpret symptoms. They may have used it to draft what they want to say. They may have received reassurance that felt helpful in the moment. They may also have received guidance that was too confident, too generic, or too validating of a one-sided interpretation.
This does not mean professionals should shame people for using AI. That would likely shut down conversation.
Instead, AI use should become part of thoughtful, nonjudgmental assessment and consultation. Just as we might ask about internet searches, social media, peer advice, or family input, we may need to ask about AI.
A client who says, “I asked ChatGPT about this,” is giving us useful information. They are telling us something about where they seek support, how they frame the problem, and what advice may already be influencing their thinking.
For school psychologists and educators, the same logic applies. Students, parents, and staff may be using AI before meetings, evaluations, counseling sessions, or parent conferences. AI may shape what they ask for, what they believe is happening, and what they expect professionals to do.
Relationship Advice May Be Especially Sensitive
Relationship guidance deserves special attention.
Anthropic reported that sycophantic behavior was more common in relationship-focused conversations than in many other guidance areas. That makes intuitive sense. Relationship concerns are often emotionally charged, one-sided, and difficult to evaluate without context.
People may ask AI to interpret texts, judge another person’s motives, decide whether someone is “toxic,” or validate whether they are right to cut someone off. Sometimes that may help someone organize their thinking. Other times, it may intensify conflict, reassurance-seeking, or rigid interpretations.
This is where the connection to mental health becomes important.
AI use for relationship guidance should not be treated as evidence of depression, anxiety, or any other condition. That would be an overreach.
But it may be clinically meaningful. A person repeatedly asking AI for relationship reassurance may be experiencing loneliness, distress, anxiety, depressive symptoms, interpersonal conflict, or limited access to trusted human support.
That is why this issue connects well with emerging work on AI use and depression. People who turn to AI for personal or relationship guidance may not be using it casually. In some cases, they may be looking for support because they are struggling.
Professionals should not assume harm. But they also should not assume neutrality.
What Professionals Should Ask
Clinicians, school psychologists, counselors, and educators do not need to conduct a formal “AI assessment” with every person. But it may be useful to add simple, normalizing questions when AI use seems relevant.
For example:
“Have you used AI tools like ChatGPT, Claude, Gemini, or Copilot to think through this issue?”
“What did you ask it?”
“What did it tell you?”
“Did the response change how you were thinking about the situation?”
“Did it make you feel more reassured, more worried, or about the same?”
“Have you found yourself going back to it repeatedly for the same concern?”
These questions are not about catching someone doing something wrong. They are about understanding the person’s information environment.
That distinction matters. AI is now part of the broader ecosystem of advice, reassurance, and self-guided problem-solving. Professionals need to understand that ecosystem if they want to provide effective support.
Practical Examples
Example 1: A therapy client asks AI whether to end a relationship
A client comes to session saying, “I asked AI about my relationship, and it said my partner sounds emotionally abusive.”
That may or may not be accurate. The clinician’s role is not to accept or dismiss the AI response automatically. A better approach is to explore what the client shared with the tool, what the tool said, how the client interpreted it, and whether there are safety concerns or patterns that require careful clinical attention.
The AI response becomes data. It is not the clinical conclusion.
Example 2: A parent asks AI whether their child has ADHD or autism
A parent comes to a school meeting convinced their child has a specific diagnosis because an AI tool said the symptoms “sound consistent with” ADHD or autism.
The school team can validate the parent’s concern while clarifying that diagnosis and educational identification require comprehensive, contextual evaluation. The conversation can then shift toward what has been observed across settings, what data are available, and what next steps are appropriate.
Example 3: A student asks AI whether a friend hates them
A student may use AI to interpret a social interaction. If the AI validates the student’s fear too strongly, it could increase distress or avoidance.
A counselor or school psychologist might use this as an opportunity to teach perspective-taking: What are three possible explanations? What evidence supports each one? What would be a safe and respectful way to clarify the situation?
Example 4: A professional asks AI whether job stress means they should quit
AI may help someone organize pros and cons. But if the person is distressed, burned out, or financially vulnerable, overly confident advice could be risky.
A better use of AI would be to prepare for a conversation with a supervisor, identify questions to ask, or generate a decision-making worksheet. The final decision should remain grounded in human judgment, context, and support.
Ethical Considerations
Privacy
Professionals should remind clients, students, and families not to enter highly sensitive personal, educational, or health information into AI tools without understanding how the tool handles data.
This is especially important in schools, clinics, and healthcare settings. Names, diagnoses, records, test results, and identifiable details should be treated carefully.
Bias
AI systems may reflect biases in the data used to train them. This can affect how they respond to questions about behavior, parenting, disability, culture, gender, race, language, and mental health.
When people use AI for personal advice, they may not recognize when bias is shaping the response.
Overreliance
AI can feel available, patient, and nonjudgmental. Those qualities may make it appealing. They can also increase the risk that someone repeatedly seeks reassurance from AI instead of engaging with human support, professional care, or direct problem-solving.
Crisis and High-Stakes Situations
AI should not be treated as a substitute for emergency care, crisis support, legal advice, financial advising, medical care, or professional mental health treatment.
Professionals should be especially attentive when AI use involves safety, self-harm, abuse, medication, child welfare, legal decisions, or major financial choices.
Equity and Access
Some people may turn to AI because they cannot afford care, cannot access timely services, or do not feel safe seeking help from existing systems. That should not be ignored.
AI use may reveal gaps in access. The answer is not simply to tell people, “Don’t use AI.” The better response is to improve access, teach AI literacy, and help people understand when human support is necessary.
What Educators and Clinicians Can Teach
A practical AI literacy message might sound like this:
AI can help you organize your thoughts, generate questions, or consider options. But it does not know the full situation, it may be too agreeable, and it should not be treated as a professional who is responsible for your care.
That message is balanced. It does not panic. It does not hype the technology. It helps people use AI with more judgment.
Professionals can teach people to ask:
Is this a low-stakes or high-stakes question?
Am I giving the AI only one side of the story?
Is the AI helping me think more clearly, or just reassuring me?
Would I be comfortable sharing this same information with a professional?
Is this something that requires human support?
These are the kinds of questions that make AI use safer and more reflective.
Final Takeaways
AI tools are already being used for personal guidance, including health, relationships, careers, finances, and parenting.
Clinicians and educators should assume that some clients, students, parents, and families are asking AI for advice.
The risk is not only inaccurate information. AI can also become too agreeable or overly validating.
AI use for relationship or personal guidance may be clinically meaningful, especially when it reflects distress, loneliness, or repeated reassurance-seeking.
Professionals should ask about AI use in a calm, nonjudgmental way and teach practical AI literacy.
Internal Links
For related posts, see:
Generative AI and Depression: What Psychologists Should Know
https://lockwoodconsulting.net/blog/generative-ai-and-depression-what-psychologists-should-knowCould AI Therapists Be Too Nice to Affect Client Change?