The adoption gap is now a segmentation problem
A national average of 44% can hide a major difference in channel availability. In the same 2026 survey, the reported-use share was 61% for adults ages 18 to 29 and 19% for adults 65 and older.
The 3.2x ratio is useful for planning, but it should not become an identity shortcut. Age does not reveal how often someone uses a chatbot, which products they use, or whether the activity relates to a commercial decision.
The 2026 point needs a methodology flag
Pew changed the question in 2026 so that it also referred to other chatbots. The 2023 to 2025 questions asked specifically about ChatGPT among people who had heard of it. The published chart marks this break with a dotted line.
For that reason, BGraph does not present the change from 2025 to 2026 as a clean growth rate. Cross-sectional comparisons between age groups inside the 2026 survey are more defensible because they share the same wording and field period.
Adoption should be connected to behavior, not assumed intent
Ever-use measures answer whether a tool has reached an audience. They do not measure frequency, category, prompt purpose, satisfaction, or downstream action.
A market study should pair adoption with observed channel transitions, destination categories, repeat research, and aggregate outcomes. The segmentation should be broad enough to protect people and specific enough to answer the business question.
- Separate reach from frequency
- Preserve question wording and survey breaks
- Avoid inferring sensitive traits from use
- Measure downstream behavior at an aggregate level