by Valerie Harkins Lindsay, Executive Director, Maternity Housing Coalition

A resident comes into your office upset. Her boyfriend is controlling. At least, she thinks he might be. She spent the last hour describing their relationship to an artificial intelligence chatbot, copied and pasted some of his text messages, and asked whether his behavior was abusive. The chatbot gave her an answer.

Or perhaps the question was about her pregnancy. She has been nauseated for three days and is worried about the baby, so instead of calling her doctor, she describes her symptoms to AI. Maybe she asks whether she should take a medication, whether her house manager is being unfair, whether your program rules are reasonable, whether she should reconcile with her mother, or whether she should move out.

By the time she comes to you, she has already had a long conversation about the problem. You just were not part of it.

This is quickly becoming a new reality of case management.

The Invisible Voice in the Room

Artificial intelligence is becoming an ordinary source of information and advice. In a 2026 Pew Research Center survey, 22 percent of American adults reported getting health information from AI chatbots at least sometimes. A 2025 study of adolescents and young adults found that about one in eight had used generative AI for mental health advice. Among 18 to 21 year olds, the number was closer to one in five. More than 93 percent of those users said they found the advice helpful.

For maternity homes, this matters. AI is available at 2:00 in the morning. It does not look tired, interrupt, or appear shocked by the question. A resident can explain the same problem seventeen different ways and the chatbot will continue to respond. Perhaps most importantly, it gives an answer.

That combination makes AI incredibly useful, but also incredibly influential.

It would be easy to dismiss this as the newest version of residents searching their symptoms online. AI is different. A search engine gives a person places to look. Generative AI has a conversation. It can take a long, emotional account of an argument and organize it into patterns. It responds in a warm and often affirming tone. It can explain a medical term at a fifth grade reading level and then explain it again using an analogy.

The problem is not that AI is always wrong. In fact, it can be remarkably helpful. A resident can use it to develop questions for her obstetrician, prepare for a job interview, understand a complicated document, or organize a chaotic list of symptoms into a timeline for her physician.

The problem is that she may not know when it is wrong.

The Answer Depends on the Question

AI only knows the information it is given, and the way a question is framed matters.

Imagine a resident asking, “My maternity home is threatening to take away my privileges because I was late for curfew. They know I have trauma and ADHD. Is this controlling?”

Now imagine her asking, “I live in a maternity home with a curfew I agreed to when I entered the program. I have repeatedly returned late, and staff gave me a consequence outlined in the resident handbook. I have trauma and ADHD and struggle with time management. How should I handle this?”

Same resident. Same consequence. Very different question.

Of course, human beings do this too. We tell stories from our own perspective, emphasize certain details, and sometimes leave out information that does not fit our interpretation of what happened. Good case management introduces context, history, accountability, and alternative ways of understanding a situation. AI cannot walk down the hall and ask the house manager what happened.

This means a resident may now arrive at a meeting with a fully developed interpretation of an event that has already been affirmed, organized, and given psychological language by a chatbot. She may have been told to “set a boundary.” She may believe a normal conflict is “toxic.” She may have a list of symptoms associated with a diagnosis or a script for confronting staff.

Our response should not be defensiveness or an eye roll. We need to become curious.

A New Case Management Question

Maternity homes routinely ask residents about their support systems. Who do you call when you are upset? Who helps you make important decisions? Who do you trust?

Perhaps it is time to add another question: “Do you ever use AI to help you think through personal, health, pregnancy, or relationship questions?”

A resident who occasionally uses AI to write a professional email is different from a resident who spends two hours every night processing her emotions with a chatbot. Staff need to understand how these tools are being used.

Ask what kinds of questions she brings to AI. Has she ever made a decision based on an answer it gave her? How does she decide whether the information is accurate? Has she shared medical records, test results, text messages, or other personal information with it?

These are not interrogation questions. They are digital literacy questions, and digital literacy has changed dramatically in the last few years.

Teach Verification, Not Fear

I believe residents need three simple questions when using AI.

First, what is the risk if this answer is wrong? There is a major difference between asking AI to create a grocery list and asking whether vaginal bleeding at 28 weeks requires medical attention. The higher the consequence of a wrong answer, the less appropriate it is to rely on AI alone.

Second, can I verify this somewhere else? Health questions may need a physician, midwife, nurse, or pharmacist. Legal questions may need an attorney or legal aid organization. Questions about the maternity home should be checked against the resident handbook and discussed with program staff. AI can help someone develop a better question for an expert, but it should not always replace the expert.

Finally, what information did I give the AI? Intensely personal disclosure to a chatbot is still digital disclosure. Medical documents, identification, court paperwork, intimate photographs, and screenshots containing another person’s private information should not be uploaded casually.

Your Staff Are Using It Too

We cannot have an honest conversation about residents and AI without discussing employees. Your case managers are probably using AI too. Someone on your team has likely asked it to rewrite an email, develop a life skills lesson, summarize a document, or help think through a difficult situation.

The question is whether your organization has established boundaries.

Can a staff member paste a resident’s case note into an AI tool and ask it to summarize the case? Can she upload a psychological evaluation? What about copying a resident’s angry text message and asking, “What mental illness does this sound like?” Can a resident’s name, date of birth, or medical information be included in a prompt?

If your employee handbook is silent on this, silence is not a policy. It is simply an unexamined risk.

Maternity homes do not need a 47 page artificial intelligence manual. Every program should, however, establish basic boundaries around confidential information, resident records, medical information, identifying data, and the use of AI in case management decisions.

AI Literacy Is Becoming a Life Skill

Our work in maternity housing has always required us to look beyond the immediate crisis. We are not simply trying to help a woman survive Tuesday. We are trying to help her develop the judgment, skills, relationships, and internal resources she will need when there is no staff member down the hall.

The women we serve will use artificial intelligence to apply for jobs, interpret medical information, navigate relationships, make financial decisions, and parent their children. We can pretend they will not, or we can help them become wise users of a powerful tool.

This does not require us to become technology experts. It requires us to do what good maternity homes have always done: pay attention to the world our residents actually live in, ask better questions, teach discernment, and refuse to confuse access to information with wisdom.

The next time a resident walks into your office with a surprisingly polished explanation of why her boyfriend is a narcissist, her medication is dangerous, or your curfew is psychologically harmful, ask one more question: “Who have you talked to about this?”

If she says, “AI,” do not roll your eyes. The conversation has already started—make sure your program is ready to join it!