State of the Prompt

Who asks about what

Pick a subject and a way of describing people, and see how often each group brings that subject to ChatGPT. Every figure is an aggregate over thousands of conversations.

Compare by
Measure

Everyone

–

Highest

–

Lowest

–

Gap

–

highest against lowest group

In plain words

Written from the published totals

GroupPeopleConversations

Every subject at once

Half as oftenTwice as often as everyone

What the shares mean

Share of conversations counts every conversation, so heavy users count for more. Share of people is the percentage of people in a group who had at least one conversation mainly about the subject. A dash marks a figure withheld because too few conversations, or a handful of people, stand behind it.

How subjects were coded

A language model read the person’s side of each conversation and named its main subject from a list of 28, or marked it as unclear or not on the list. Creative writing was added to the list later, so its share is a lower bound.

Who is in the data

65,689 adults on a US research panel, counting the conversations they had at 18 or older, July 2025 to June 2026. Age groups use each person’s age in July 2026. The panel is not a random sample of the country, so compare groups rather than reading any figure as national. More about the data.