OpenAI data shows users shifting from asking questions to doing work

Remember when ChatGPT was mostly just a fancy parlor trick where you asked it trivia or told it to write a poem about pizza? Those days are officially over.

OpenAI just dropped a massive dataset on its Signals platform through its Economic Research Team, breaking down how over 1 billion people actually use ChatGPT Free, Go, Plus, and Pro accounts across different countries.

The verdict? The world is rapidly moving from asking questions to executing tasks.

Getting down to business

The clearest shift is happening during work hours. Users in professional settings are now more than twice as likely to use ChatGPT to complete a task or create an output—such as writing code, editing drafts, or conducting data analysis—compared to their personal time.

Outside of work, ChatGPT still looks like an exploratory research tool where people ask questions and look for explanations. But inside the office, it is being treated like a digital co-worker expected to produce deliverables.

Meanwhile, multimedia is on a tear. Following the release of ChatGPT Images 2.0 in April 2026, multimedia interactions jumped to 7.8% of all global messages. In countries like Brazil and Colombia, multimedia now accounts for more than one in ten queries.

The Global South and over-35s are catching up

The demographic divide is also closing fast. While North America and Europe dominated the early days, Latin America, Oceania, and Africa are surging in per-capita usage.

In Q2 2026, Peru, Uruguay, and Costa Rica saw the fastest jumps on the global leaderboard as adoption spread well beyond early-adopter hubs.

At the same time, older demographics are moving in. Globally, users aged 35 and older now account for a 5% higher share of messages than they did a year ago.

In France and Czechia, the share of messages coming from the over-35 crowd jumped by more than 10 percentage points over the past 12 months. Nearly three-quarters of European countries saw above-average growth in this demographic, whereas Southeast Asian hubs like Singapore experienced much milder increases.

Why it matters

We think this transition from querying to doing marks the true maturation of generative AI into an operational layer for knowledge work.

For everyday professionals, the takeaway is simple: if you are still just treating ChatGPT as a smarter Google, you are missing the boat. The real efficiency gains come from delegating end-to-end deliverables like code refactoring and data synthesis.

For the tech industry, OpenAI is subtly signaling where the next battleground lies. By proving that 1 billion users want execution rather than conversation, it pressures competitors like Anthropic and Google to focus on agency and structured outputs rather than pure chat interfaces.

The obvious catch here is accountability. When users transition from gathering advice to deploying AI-generated code and business analyses, hallucinations stop being minor embarrassments and become enterprise liabilities. Making sure these task outputs remain dependable is going to be the hardest hurdle ahead.