Hello,
I’m a web developer and frequent ChatGPT user, and I wanted to share a UX idea that could significantly reduce friction in conversational interactions.
ChatGPT often asks questions where the answer is clearly discrete or limited to a few options, for example:
-
“Do you want a short or detailed explanation?”
-
“Should I continue with option A or option B?”
-
“Would you like code, a step-by-step plan, or a summary?”
In these cases, users currently need to manually type the option, even though the assistant already understands the possible choices.
A potential improvement would be allowing the assistant to optionally return structured reply options, which the client UI could render as native buttons or chips (depending on the platform: web, Android, iOS, etc.).
Example concept:
{
"message": "Which format would you prefer?",
"choices": [
{ "label": "Short answer", "send": "Give a short answer" },
{ "label": "Detailed explanation", "send": "Give a detailed explanation" },
{ "label": "With examples", "send": "Explain with examples" }
]
}
The client application could then render these as:
-
buttons in web
-
chips or quick replies on mobile
-
selectable actions in desktop UI
Benefits:
• reduces typing friction, especially on mobile
• speeds up common decision points in conversations
• preserves free-text input (users can still type anything they want)
• leverages the model’s ability to recognize when a question has discrete options
Importantly, this would not require the assistant to generate raw HTML. Instead, the model could return structured metadata, and the client would render the appropriate UI element for each platform.
I believe this hybrid approach (free chat + contextual UI actions) could improve usability without restricting the open-ended nature of the conversation.
Thank you for the great work on ChatGPT.
Best regards.