How people actually use ChatGPT

When we talk about AI, the focus often lands on workplace productivity, coding, or replacing jobs. But a new study from OpenAI shows a very different story: most people are using ChatGPT outside of work. And that shift is worth paying attention to.

AI Summary

• A new study reveals a key shift in ChatGPT usage: despite initial workplace focus, personal tasks now dominate (70% vs. 30% work-related prompts), indicating a move towards consumer integration rather than pure work-tool application. This trend is significant as it shows AI's everyday utility expanding beyond traditional job-related contexts.
• While coding and data analysis were initially hypothesized as central uses, the study debunks these strongly. Coding accounted for a mere 4.2% of prompts, and data analysis a surprisingly low 0.4%. This highlights the current prevalence of ChatGPT for "lighter-lift" tasks like planning, drafting, brainstorming, and learning, rather than large-scale technical functions.
• The dominant narrative of AI as a workplace rescuer is thus nuanced. The study emphasizes AI's role as a "co-pilot" or helpful companion, particularly outside of formal work environments. AI's true impact might be felt most in everyday life, enriching individual productivity and potentially reshaping various sectors organically rather than through a sweeping job-replacement scenario.

Personal vs. work use

The biggest finding is simple: most people use ChatGPT for personal tasks, not work. Around 30% of prompts were work related, while 70% were non-work. That share has shifted over time — in June 2024, work made up nearly half, but by June 2025 it dropped to just over a quarter. Meanwhile, overall usage grew from 451 million prompts to more than 2.6 billion.

This suggests ChatGPT is becoming more of a consumer app than a workplace tool. It's part of daily life for many people, whether for writing, planning, or exploring ideas outside of work.

Coding isn't the main use

There's been a lot of talk about AI replacing programmers. The study paints a different picture. Only 4.2% of ChatGPT prompts were about coding.

Even when looking at Claude, another AI tool with a reputation for technical tasks, the numbers aren't overwhelming. One study found 37% of its prompts were related to computing or math. That's higher than ChatGPT, but still far from a majority.

The conclusion: coding is important, but not the dominant use case for AI chat. One caveat noted in the study is that much of AI assisted programming has shifted to specialized tools (such as Cursor) that use the API — meaning this activity happens outside of ChatGPT.

Therapy and advice

Another common idea is that AI is mainly used for therapy or emotional support. In practice, just 1.9% of ChatGPT messages were about that. So while some people do use it in this way, it's not a central use case.

Instead, most usage falls into lighter, practical categories — planning, drafting, brainstorming, and learning. This reflects AI's role as a helpful companion, not a replacement for professional support.

Data and analytics: still small

One striking detail: only 0.4% of prompts were about data analysis. That's surprisingly low given how often companies highlight “AI-powered analytics.”

It suggests that while vendors talk up these capabilities, interest is much smaller. Most users aren't turning to AI for heavy data work. They're focusing on simpler, more direct tasks.

This may change in the future, though. For example, as companies start using small language models (SLMs) more widely. These models can be trained for specific tasks, run on lighter infrastructure, and keep data more secure.

The bigger picture

The study concludes that AI is most effective as a co-worker or a co-pilot. It helps educated users solve problems faster, rather than replacing them. The benefits are real, but uneven.

Instead of sweeping job replacement, we see something subtler: AI woven into everyday life, used mostly outside work, and in ways that improve productivity for some groups more than others.

Conclusion

The lesson is this: AI adoption is not just about the office or the factory floor. It's happening at the kitchen table, in classrooms, and in people's side projects. That may be where its biggest impact shows up first.

This also aligns with leading AI figures who are trying to cool down the hype and warn about a possible “AI bubble.”