From Prompt Curious to Prompt Power User: Who’s Actually Talking to AI?

Awareness of a generative AI tool is only half the battle – getting people to use it is the other half. We wanted to better understand which platforms people are using, so we asked them in our most recent round of research-on-research.

It’s important to note that this question was only asked to respondents who had previously said they were aware of these platforms.

Let’s dive into what we uncovered.

Overall

Across all respondents, ChatGPT stands out with 49% of people reporting usage in the past 30 days. Gemini is next at 33%, followed by Copilot at 22%. Smaller platforms such as Grok, Claude, and Perplexity remain under 10%, with Duet at 3%. Despite this growth, 30% of respondents still reported using none of these tools, showing that a significant portion of the population has yet to integrate generative AI tools into their daily routines.

Wave-Over-Wave

Comparing the data across the last five waves of our research-on-research, we can see that usage has steadily increased. ChatGPT only had 29% usage in Wave 1, rising to 49% in Wave 5. Gemini shows an even sharper climb, jumping from 9% in Wave 1 to 33% in Wave 5. Copilot also grew consistently, from 9% to 22% across the same period. 

Gender

Comparing the data by gender, men are more likely than women to report using generative AI. Fifty-seven percent of men used ChatGPT in the past 30 days compared to 41% of women. A similar gap appears for Gemini (41% vs. 24%) and Copilot (28% vs. 16%). Women are also more likely to report no usage at all, with 37% selecting “none of the above” compared to 22% of men. The divide is especially visible for smaller platforms, where adoption remains limited among women.

Age

Age seems to have a strong correlation with usage. ChatGPT usage peaks among the youngest adults, with 65% of those aged 18–24 and 61% of those 25–34 reporting recent use. Usage declines steadily with age, falling to 23% among those 65 and older. Gemini follows a similar path, with the highest adoption among middle-aged adults (39% of 25–34 and 43% of 35–44) and much lower use among older groups.

Younger adults also drive the adoption of newer platforms. Around 9% of respondents under 45 used Claude in the past month, compared to nearly none among older adults.

Income

Higher income households show greater usage of AI tools. ChatGPT usage jumps from 30% among those earning under $20,000 to 67% among those earning $100,000 or more. Gemini rises from 25% in the lowest income bracket to 43% in the highest. Copilot follows the same trend, ranging from 14% to 34%. Lower-income households are also far more likely to report no use, with 40% of the under $20,000 group selecting “none of the above” compared to just 18% among the highest earners. The connection between income and usage is clear across all platforms.

Political Affiliation

Generative AI usage also differs depending on political affiliation. Republicans report the highest usage of ChatGPT at 55%, followed by Democrats at 50% and Independents at 43%. Gemini use is also higher among Republicans and Democrats (37% and 36%) compared to Independents and Others. Copilot and Claude follow the same pattern, with Republicans and Democrats ahead of other groups.

Ethnicity

We found that ethnicity also impacts generative AI usage. ChatGPT is most used by Asian (65%), African American (55%), and Hispanic/Latino (52%) respondents, compared to 45% of Caucasians. Gemini adoption is higher among minority groups as well, reaching 39% among Hispanic/Latino respondents compared to 31% of Caucasians.

Claude sees its highest usage among Hispanic/Latino respondents at 11%, compared to 6% among Caucasians. Meanwhile, “None of the above” is most common among Caucasians (33%), while minority groups report lower levels of non-use.

Panel

Similar usage variations are also present at the panel level. Panel H reports the highest usage overall, with 65% using ChatGPT and 44% using Gemini. In contrast, Panel O is lowest, with just 35% using ChatGPT and 29% using Gemini. That 30-point gap in ChatGPT usage between Panels H and O is the widest among all tools. Copilot also shows a strong split, with usage reaching 28% in Panels E and H but dropping to only 15% in Panel C. Those reporting no usage are also unevenly distributed, from just 20% in Panel H to 41% in Panel O.

These sharp contrasts illustrate how panel composition can affect results, making strategically blending sample the best way to balance these extremes.

Find out how Strategic Sample Blending can help you, request a free consultation.