This topic is related to AI in developmental support and social care, but I would like to focus specifically on the user-side experience: how different AI model interaction styles affect neurodivergent users.
I’d also like this to be a place where people can share practical experiences, not only abstract opinions.
For example: models that were difficult to use, models that became easier after an update, custom instructions that helped, or interaction patterns that made things better or worse.
If enough people have similar experiences, I hope this thread may also become a useful small sample of user-side feedback for future AI interaction design, accessibility, and model behavior improvements.
I’d like to ask something to other neurodivergent users.
I’m also someone with ADHD, diagnosed by a medical institution in Japan.
I’m interested in this topic not as a specialist, but from the perspective of a user who works with GPT in daily life.
I regularly use GPT for thinking, writing, dialogue, and product design.
As I continued using it, I noticed that even within GPT, different models can feel very different.
The models I had difficulty with were not necessarily hard to use because they gave wrong answers.
Rather, the way they interacted sometimes did not match my cognitive style.
For example:
-
pushing too quickly toward conclusions
-
using too much summarization or reassurance
-
repeatedly ending conversations with binary questions
-
moving into emotionally deep areas before I was ready
-
making me feel as if I was being treated as an “average user” rather than as a person with my own specific context
For me, a good AI interaction needs pacing.
It needs to be able to hold thoughts in an unfinished state.
It needs not to rush me into making choices.
And just because I am able to answer a question does not mean the model should move into the depth that the question opens.
I feel that this kind of adjustment in pace and depth is very important.
Has anyone else had similar experiences?
I’d be interested to hear if certain AI model interaction styles have helped you, or on the other hand, made things harder.