top of page

ChatGPT for Therapists: Why Your Expertise Matters Most at 7:42 P.M.

2d
5 min read
Stacks of printed drafts spread across a desk during the development of an AI-supported resource

As a therapist, have you ever worked with a parent in a situation like this? She actively participates in sessions, she takes notes, she tries to apply every single technique at home week after week, and something just isn't sticking. She understands all the techniques for working with her child with Big Emotions. Her neurodivergent child has a hard time stopping craft projects and moving into bedtime routine. Some nights this transition is manageable. Other nights, the five-minute warning comes and the whole evening starts to snowball like an avalanche. You have talked at length in sessions about what tends to happen before her child becomes overwhelmed. You have practiced keeping the words minimal, giving advance notice, using predictable steps, and limiting how many choices are offered at once. Sitting with you in the session, she can explain the plan back to you. But, she is in no way prepared at the time of the avalanche between crafttime and bedtime routine. Each time it sneaks up on her, and she freezes. Do we know why?


When it's already 7:42 p.m. 18 minutes before bedtime. Dinner ran late. Craft time was shorted. Her child finally settled into a favorite activity and does not want to stop. She gives the five-minute warning and hears, “No!” A few minutes later she tries again. Her child is getting louder. A sibling needs help at the same time. The PJ's are still in the dryer. Toothbrushing has not even started. The parent can feel herself adding more words because nothing seems to be working. And at this point she may be getting frustrated. Perhaps, even her nervous system has become dysregulated. She knows there was something you talked about for this, but she cannot get to it fast enough right now.


She may actually be the focus of the learning need at bedtime routine, its not necessarily a problem with remembering that she is suppose to stay calm... but a skill gap where the parent knows how to regulate herself during the episode so she can continue through the routine. She may not need more information about transitions. She may need to be able to use what she already learned during the part of the evening when it is hardest to access.


Now, let's say you want to give her something to take home, so you open ChatGPT and write a thoughtful prompt. You ask for a neurodiversity-affirming parent guide about transitions and changes in routine. You ask it to explain why transitions can be difficult, include a four-step sequence such as Preview, Countdown, Choice, Transition, give parents language they can use, suggest visual supports, keep the tone compassionate, and make the resource easy for a tired parent to scan.




It is a strong prompt, and ChatGPT gives you a polished handout. The headings are clear. The language is warm. There is a visual sequence, sample phrases, a timer, a first-then support, and several ideas the family could try. Nothing immediately looks wrong. Your client could ask ChatGPT for something very similar herself.


But you know more about this parent than the prompt does. You know what happens at 7:42 p.m. when the five-minute warning does not help. You know which strategies the family has already tried, where the parent tends to get stuck, what the child does when the transition becomes harder, and what the parent can realistically remember when both of them are overwhelmed. That changes what this parent may actually need from you.


She may already understand transitions. She may already know the four steps on the handout. The gap may be that she can explain the strategy in your office but can't use it when bedtime is already failing. Before creating another resource in ChatGPT, you need to know what she can do now, where the skill breaks down, and what would help her get from one to the other.


If those details never make it into the ChatGPT thread, the model still produces an answer for both the therapist or the parent, even if the answer is incorrect for you. One response may emphasize using visual schedules. Another may add sensory ideas. Another may give the parent six scripts, five tools, and several more strategies to try. All of it can sound reasonable.


Then you are back inside the draft trying to make it fit. You delete suggestions the family has already tried. You rewrite examples. You end up removing a reward system that was never part of your work with this family. You shorten the scripts. You bring back the phrase the parent practiced in session. You reduce the number of choices. You change the bedtime example because it assumes a routine this family does not have. The AI made the handout fast. You are still supplying the professional thinking that makes the handout useful.


Generative AI makes it easier to create worksheets, guides, scripts, checklists, and visual supports. Your clients have access to those tools too. A parent can ask ChatGPT how to help a neurodivergent child with bedtime transitions and receive a long list of reasonable ideas.


Your expertise and value as a professional though, is not in competing with that list that ChatGPT can give them. You know which part belongs with this child. You know what has already been tried. You know where the routine breaks down. You know what the parent can realistically remember when everyone is tired. You know what they practiced together last week. You know whether the parent needs more information or needs help using what they already know and regulating themselves.


At a Private AI Pairing Table™, Bump Bench begins with one defined learning need from your own practice. In this example, the work would not begin with “create a handout about bedtime transitions.” It would begin with the parent who needs to remember and use the transition approach when she and her child becomes overwhelmed during the transition from evening play into bedtime.


You bring what you know about the child, the caregiver, what has already been tried, where the routine becomes difficult, and what the caregiver needs to be able to do. Instructional design helps shape the resource and the written AI instructions around that learning need. Then you test the work in your own ChatGPT or Claude account and decide what fits. Your professional judgment stays yours.


Your client can ask ChatGPT for a bedtime routine. Your expertise matters more at 7:42 p.m., when the five-minute warning did not help, their child is already overwhelmed, bedtime still has to happen, and the parent is trying to remember what to do next.

 
 
 

Comments


bottom of page