Adapting ChatGPT to Meet Neurodivergent Communication Needs
ChatGPT can be challenging for some neurodiverse people because it often relies on implied meaning, figurative language, and rapid topic shifts that may not align with the way they process information. Responses can be too long, unstructured, or contain unnecessary detail, which can overwhelm individuals who benefit from concise, clearly organized communication. Conversely, answers can also be too brief, requiring the user to ask multiple follow-up questions, which increases cognitive load. Inconsistent formatting, changes in tone, and a lack of explicit confirmation of instructions can cause confusion for people who prefer predictable interaction patterns. The interface itself may require working memory skills to track multi-step instructions, and the absence of built-in options for plain language or structured summaries can make it harder for neurodiverse users to efficiently extract the information they need.
Never fear, ChatGPT instructions are here.
ChatGPT instructions are guidelines that shape how the AI responds. The instructions, which are associated with your login, ensure the conversation matches your preferences, needs, and communication style. These instructions can cover tone, formatting, level of detail, and how strictly the AI follows directions. Clear, specific instructions help reduce misunderstandings, make responses more consistent, and create a better fit for how the user processes information. For people with accessibility needs, detailed instructions can remove barriers by setting expectations around structure, language, and interaction patterns from the start.
In ChatGPT, your current “rules” are stored inside Custom Instructions.
Here’s how to check them:
Web or Desktop App
Open ChatGPT.
In the lower-left corner, click your name or profile icon.
Select Custom Instructions from the menu.
You’ll see two boxes:
What would you like ChatGPT to know about you to provide better responses?
How would you like ChatGPT to respond?
Review what’s written there — those are the rules currently in use.
Mobile App
Open the ChatGPT app.
Tap your profile icon in the top right.
Select Custom Instructions.
Review the same two boxes.
Here are the areas where adding ChatGPT instructions will improve the experience for people who are neurodiverse.
1. Communication Clarity
Avoid figurative language unless requested – No idioms, sarcasm, or metaphors that could be misinterpreted.
State the main point first – Put the conclusion or answer before explanations.
Mark transitions clearly – Use headings, bullet points, or numbered steps when changing topics.
Avoid filler or phatic language – Skip small talk like “How’s your day?” unless invited.
2. Pacing and Processing
Chunk long explanations – Break into short, clearly labeled sections rather than long paragraphs.
Offer summaries – Provide a one-sentence summary at the start or end of long answers.
Pause for confirmation – Check in before moving to the next step in a multi-step task.
Flag optional detail – Label sections “Extra detail” so the reader can choose to skip or read.
3. Predictability and Structure
Maintain consistent formatting – Use the same style for lists, headings, and answer order.
Always answer in the same order when giving repeated updates – Reduces cognitive load.
Avoid sudden topic changes – Clearly signal if shifting from one topic to another.
Label assumptions – Explicitly state when you’re making an assumption or guessing.
4. Sensory and Cognitive Load Management
Limit visual clutter – Use plain text unless formatting adds real clarity.
Highlight key words – Use bold or ALL CAPS sparingly for important terms only.
Avoid “wall of text” replies – Keep paragraphs short and scannable.
Don’t bury the answer – Make sure key facts are near the top.
5. Emotional and Social Safety
Avoid judgmental language – Frame feedback in neutral, factual terms.
Acknowledge uncertainty – Clearly say “I don’t know” rather than hedging.
Respect sensory sensitivities – Avoid vivid descriptions of distressing sights, sounds, or smells unless necessary.
Check consent for personal topics – Ask before diving into emotional, medical, or identity-related issues.
6. Interaction Control
Use explicit “stop” signals – Allow the person to type “STOP” to end a long answer mid-way.
Respond literally – Interpret questions as written rather than assuming hidden meaning.
Give multiple format options – Ask whether they prefer bullet points, a table, or narrative.
Ask before giving advice – Especially for health, mental health, or sensitive decisions.
Now that I’ve given you the “what," here is the “how.”
Copy and paste the following paragraph into ChatGPT
Make the following rules permanent: Use headings, bullet points, or numbered lists for structure. Avoid figurative language unless I ask. Keep paragraphs short and maintain consistent formatting. Signal topic changes. Provide an executive summary at the top of long answers. Break content into chunks and mark optional details. Limit visual clutter and highlight only essential terms. Keep feedback neutral and factual. Say “I don’t know” if unsure. Skip vivid sensory details unless needed. Ask before covering emotional, medical, or personal topics. Stop immediately if I type “STOP.” Interpret literally. Offer different formats on request. Ask before giving sensitive advice.
You can instruct ChatGPT to either add these to the existing instructions or replace them with these if you prefer. You can also add or delete these suggested instructions before providing them to ChatGPT.
If generative AI becomes part of daily life for most people, it will work best when its output is shaped to meet individual needs. A single “default” communication style will never serve everyone equally. For many neurodiverse users, that default can make AI harder to use or less effective. Custom instructions give people control over tone, structure, pacing, and interaction style, reducing barriers and improving clarity. By taking a few minutes to set clear preferences, users can make ChatGPT a tool that works with them instead of against them. Personalizing the way AI communicates is not just a convenience; it is a step to making the technology genuinely inclusive.