Open AI’s ChatGPT: How Consumers Use The Application

 In July of 2025, 700M users were sending 18 billion messages each week. The speed of adoption is unprecedented. About 10% of our global population is using the tool with the highest growth coming from emerging markets. Both adoption (i.e., number of users) and usage (i.e., how many queries sent) continue to grow. 

It is essential for anyone building GenAI tools to understand consumer behavior because it will impact the success of their own products. When new technology - devices, services, or interfaces - come to consumers, they need time to become comfortable with the technology. At first, they become aware of the new “thing” from the media or friends. Then they adopt it (i.e., trial) before becoming regular users. Once they are regular users, they develop a comfort and something even a preference to use the new technology. 

Think back to the days when tickets or mobile boarding passes were available on smartphones.  Many of us still printed out the boarding pass because we didn’t trust the process i.e., we weren’t yet comfortable with the technology. 

Here are my top take-away’s: 

• The majority of usage is still a replacement for Search or asking questions. The authors divided user activities into three categories: asking (49%), doing (40%), and expressing (11%). In the “asking” category, seeking practical guidance edges out searching for information.

• Once you look past the three aggregate categories, there is a very long tail of use cases in the workplace. (Again, there is the caveat that ChatGPT doesn’t represent all enterprise use - far from it.) Three use cases make up near half of conversations. They are: getting information (19.3%), interpreting the meaning of information for others (13.1%), and documenting/recording information (12.8%). My question is: are users escaping toil? Shifting behavior from Google to ChatGPT? Or asking the tools to do work they otherwise can’t?

• Most of the usage is consumption rather than creation. This pattern of behavior fits other new technology introductions. My hypothesis would be that the powerful creation abilities of GenAI would encourage more people to create. We may just be in a world where the majority of people are trying to filter and understand an abundance of information. That has its own implications especially if our critical thinking abilities continue to deteriorate. 

Here is what didn’t surprise me: 

• Early adopters of the application were young males. This is normal for new technology experiences.  The gender mix is more even now though usage still skews towards younger adults.

• Regular usage skews towards more educated adults in knowledge worker type jobs. The tools are capable of doing “digital things” i.e., what knowledge workers do a lot. Hopefully these folks tap into their critical thinking skills to assess and adapt the outputs.

• Most (75%) of the “Doing” is writing. Too much for this blog, but the right educator or psychologist could write books about how consumers spend most of their time passively consuming content on smartphones while their creation skills have atrophied. Even on social media platforms, a minority of users create most of the content.

• Use of the basic tools for work has declined. On one hand, enterprises have given their employees more secure tools to use while educating them on the dangers of putting confidential data into public spaces. On the other hand, 46% of employees surveyed by KPMG in a study published in 2025 admitted to uploading sensitive company information and intellectual property to public AI platforms.

• Work-specific tasks such as coding have declined as a percentage of the total. While this may seem counterintuitive to our perception of how folks are using GenAI given the news, it likely means they have shifted to other dedicated tools that entire teams may be using. 

Here is what did surprise me:

• Nearly 6% of consumers sought advice on health, fitness, beauty, or self-care. I’d be super curious to click down here to understand why. Do they want to understand an existing condition? Or do they want customized guidance (e.g., training for a 10K) that is otherwise inaccessible to them due to cost or personnel? The question is important because it could a) be a displacement of another online activity or b) be the addition of a service they can’t get in person. And then there is the trust in the answers. So much one could unpack. Teaching and tutoring (10%), How to Advice (8.5%), and Creative Ideation (4%) rounded out this category. 

• The study didn’t break out what type of tools consumers chose to use (e.g., video, image, deep reasoning). I wouldn’t be surprised if the simplest of models were good enough for most of their needs.

• While Expression (e.g., relationships, personal reflection, chatchat, roleplay) was the smallest volume of all three categories, users stated the highest satisfaction with these interactions compared to the others. I keep hoping that Mr. Zuckerberg is wrong in his vision of virtual companions. This is a trend to watch.

• Final note … I am surprised that adoption isn’t closer to 75% or 80% of the world’s population. Most adults have smartphones and access to these tools via a downloadable app or even WhatsApp. 

Context: The National Bureau of Economic Research (NBER) released a research paper titled “How People Use ChatGPT” in September 2025. The academic-style paper is 64 pages long and jam-packed with data visualizations, research methodology, and author conclusions. They not only look at a snapshot of today’s usage, but also analyze how usage has changed since ChatGPT’s first release. The study limits its analysis to consumer plans (e.g., Free, Plus, and Pro).  It is available for download at http://www.nber.org/papers/w34255.