OpenAI's First Study On ChatGPT Usage (arstechnica.com) 20
An anonymous reader quotes a report from Ars Technica: Today, OpenAI's Economic Research Team went a long way toward answering that question, on a population level, releasing a first-of-its-kind National Bureau of Economic Research working paper (in association with Harvard economist David Denning) detailing how people end up using ChatGPT across time and tasks. While other research has sought to estimate this kind of usage data using self-reported surveys, this is the first such paper with direct access to OpenAI's internal user data. As such, it gives us an unprecedented direct window into reliable usage stats for what is still the most popular application of LLMs by far. After digging through the dense 65-page paper, here are seven of the most interesting and/or surprising things we discovered about how people are using OpenAI today. Here are the seven most interesting and surprising findings from the study:
1. ChatGPT is now used by "nearly 10% of the world's adult population," up from 100 million users in early 2024 to over 700 million users in 2025. Daily traffic is about one-fifth of Google's at 2.6 billion GPT messages per day.
2. Long-term users' daily activity has plateaued since June 2025. Almost all recent growth comes from new sign-ups experimenting with ChatGPT, not from established users increasing their usage.
3. 46% of users are aged 18-25, making ChatGPT especially popular among the youngest adult cohort. Factoring in under-18 users (not counted in the study), the majority of ChatGPT users likely weren't alive in the 20th century.
4. At launch in 2022, ChatGPT was 80% male-dominated. By late 2025, the balance has shifted: 52.4% of users are now female.
5. In 2024, work vs. personal use was close to even. By mid-2025, 72% of usage is non-work related -- people are using ChatGPT more for personal, creative, and casual needs than for productivity.
6. 28% of all conversations involve writing assistance (emails, edits, translations). For work-related queries, that jumps to 42% overall, and 52% among business/management jobs. Furthermore, the report found that editing and critiquing text is more common than generating text from scratch.
7. 14.9% of work-related usage is dealt with "making decisions and solving problems." This shows people don't just use ChatGPT to do tasks -- they use it as an advisor or co-pilot to help weigh options and guide choices.
1. ChatGPT is now used by "nearly 10% of the world's adult population," up from 100 million users in early 2024 to over 700 million users in 2025. Daily traffic is about one-fifth of Google's at 2.6 billion GPT messages per day.
2. Long-term users' daily activity has plateaued since June 2025. Almost all recent growth comes from new sign-ups experimenting with ChatGPT, not from established users increasing their usage.
3. 46% of users are aged 18-25, making ChatGPT especially popular among the youngest adult cohort. Factoring in under-18 users (not counted in the study), the majority of ChatGPT users likely weren't alive in the 20th century.
4. At launch in 2022, ChatGPT was 80% male-dominated. By late 2025, the balance has shifted: 52.4% of users are now female.
5. In 2024, work vs. personal use was close to even. By mid-2025, 72% of usage is non-work related -- people are using ChatGPT more for personal, creative, and casual needs than for productivity.
6. 28% of all conversations involve writing assistance (emails, edits, translations). For work-related queries, that jumps to 42% overall, and 52% among business/management jobs. Furthermore, the report found that editing and critiquing text is more common than generating text from scratch.
7. 14.9% of work-related usage is dealt with "making decisions and solving problems." This shows people don't just use ChatGPT to do tasks -- they use it as an advisor or co-pilot to help weigh options and guide choices.
A search engine (Score:3)
provides advice too.
The 72% non-work is comical - The gimmick crowd. They vanish once the operating costs get applied.
Re: (Score:2)
The 72% non-work
That number is not credible. The prompts I write for "personal" purposes are literally indistinguishable from what a worker might make: for all ChatGPT knows I'm an auto mechanic. Location isn't a valid metric either, given WFH, mobile devices, etc. They can't possibly distinguish between work and personal to a one digit of precision. Obviously they can see whether prompts are coming from commercial accounts, paid personal accounts or free tier, but even that gets fuzzy at the low end.
Re: (Score:3)
The number is probably more credible than you'd like.
Essentially, they paired ChatGPT users with data from data brokers in a DCR and if your questions were about your known profession, then it was work related.
The data brokers likely know what color your underwear is, and that you didn't change it this morning.
Re: (Score:2)
The data brokers likely know what color your underwear is, and that you didn't change it this morning.
Do they know at what rate it changes color if I keep it several days ?
Re: (Score:2)
Hypercolor Underwear- it's a bold strategy Cotton. Let's see if it pays off for him.
Re: (Score:2)
The 72% non-work is comical - The gimmick crowd. They vanish once the operating costs get applied.
No. Some portion of those are paid users. The distribution is not given.
There is 1 person paying $200/mo, and 4 paying $20/mo in my office alone that I know of.
Re: A search engine (Score:2)
but unlike uber, who benefited from it first, this cannot even scale with infinite money
Re: (Score:2)
Want to try again?
Nothing in this world scales with infinite money, so I find that to be a pretty interesting complaint.
Re: (Score:2)
The operation costs for inference are exponentially shrinking. Model training stays up, because optimizations are used to train even better models instead of cheaper ones, but inference got down.
Currently you see rates of 9-900 (depending on how smart the model should be) decrease PER YEAR.
https://epoch.ai/data-insights... [epoch.ai]
OpenAI is still stealing everyone else's lunch. (Score:2)
Re: (Score:2)
OpenAI is still stealing everyone else's lunch.
Just as much as Google News is.
GenAI as assistance is a fair assessment (Score:2)
GenAI is just another tool.
It isn't going to make all knowledge jobs redundant. But if you want to write a story that says it does to pump the OpenAI stock it can help with you that.
It will assist knowledge jobs, perhaps marginally increase productivity and quality in the short term.
Long term it may create a dependency that stifles creativity and innovation. We don't know where new ideas come from, I expect you have to create the right conditions for them to arise in a human from practising being creative.
Checks out (Score:2)
4. At launch in 2022, ChatGPT was 80% male-dominated. By late 2025, the balance has shifted: 52.4% of users are now female.
Can talk to you, and more importantly listen to you, endlessly, never getting tired. Checks out ;)
7. 14.9% of work-related usage is dealt with "making decisions and solving problems." This shows people don't just use ChatGPT to do tasks -- they use it as an advisor or co-pilot to help weigh options and guide choices.
Even there, used wisely it can be helpful.
Ever heard of "rubber duck programming"? Try to explain what you are doing to a rubber duck. Just the act of conversing about something can give you a different perspective in thinking about it.
Re: (Score:2)
Can talk to you, and more importantly listen to you, endlessly, never getting tired. Checks out ;)
Maybe, but more probably an enormous amount of (male) usage has shifted to interaction with AI assistants via other routes. I hardly use the web interface anymore. Everything happens through applications (such as Cursor).
Re: (Score:2)
I started a small a plant nursery in the last year, and at the onset I decided I was going to use AI (I use Copilot) as much as possible to help me out.
Honestly, it's been awesome.
One of my first ideas was to use it to manage inventory, and frankly that bombed. After a month of inputting data, I started asking it to give me reports. I could ask it for the same information 3 times in a row, and get 3 wildly different answers. So, it absolutely cannot function as a database.
But it absolutely excels in assis
victms of the nyt (Score:2)
When did people forget how to write? (Score:2)
Re: (Score:2)
Hey they didn't get to executive class by being good at writing. They got there by .. I don't know how they get there. I do know that's not where I'm going.