Annie Tsai

Between January 2025 and last week, ChatGPT’s usage grew from 3.8 billion to 6 billion monthly active users representing one billion queries every single day across 188 million daily visitors. The site now ranks fifth globally, behind Google, YouTube, Facebook, and Instagram. Which raises a question worth asking: what exactly are so many people asking AI every day?

I know how I use it – technical architecture design, deep research, coding, planning and recipe building. I absolutely love using AI to cook. On Friday with my cucumber harvest I learned that I could saute cucumber, similar to a Chinese recipe using bitter melon, and it tasted out of this world. But I suspected that my usage wasn’t the norm, so I dug into usage across a few generative AI platforms to see how people are actually using these LLMs. The results say a lot about what we value, where we get stuck, and how we’re reimagining work.

Across billions of daily prompts, patterns emerge. Writing and editing sits at the top – blog posts, resumes, cover letters, emails, and endless requests to “make this sound more professional”. Close behind is idea generation: marketing hooks, slogans, wedding toast material, the kinds of creative sparks that help people break through a blank page. Summarization and simplification come next, with people feeding in transcripts, long documents, meeting notes, and academic texts, asking for the digest version. Templates and frameworks also rank highly, as people lean on AI to scaffold projects from scratch – think editorial calendars, research outlines, checklists. And increasingly, people use it to interpret data: turning sales dashboards or research reports into plain English that highlights trends. Coding and tech help has become another door in, where non-engineers debug snippets or even build from scratch. 

Learning and research questions are common too like “What was the Paris Climate Agreement?” or “Explain the Cuban Missile Crisis.” AI is now tutor, fact-checker, explainer. Students, in particular, ask for study aids like flashcards and quizzes. And then there’s the more surprising category, where people lean on AI for wellness and emotional support – stress management tips, journaling prompts, or simply a place to talk through worries.

Today, AI is being used less as a specialist tool and more as a universal productivity booster. People are offloading the low-stakes, high-friction work: first drafts, summaries and “where do I start” moments. They also reveal where humans crave judgment versus execution. We’re happy to let AI edit grammar, but we still want reassurance on whether our resume will stand out in the crowd. We’ll let AI summarize a report, but we want to decide what those findings mean for our strategy. And then there’s the surprising intimacy of it all: people trust AI with vulnerabilities and sensitive moments – anxiety before a presentation, a moment of self-doubt, a stressful work situation. In that sense, ChatGPT is less an assistant and more a mirror.

But, adoption comes with a hefty price tag. Someone can now produce in an afternoon what once took a team a week, which makes one feel like they have superpowers. Yet many of the tasks people offload to ChatGPT are the same ones entry-level workers use to cut their teeth on. If those rungs on the career ladder disappear, what replaces them? Does acceleration for the many come at the cost of fewer opportunities for the new? There isn’t an easy answer, but it’s a tension worth holding onto and proactively addressing as unemployment in that segment increases.

There are other risks too. Unless you’ve explicitly disabled it, your chats are included in AI training data and can even surface in search results. Best practice is to turn off chat history if privacy matters, never paste in personal details, passwords, or API keys, and treat prompts like a semi-public space: if you wouldn’t post it on social media, don’t paste it here. Above all, use AI as a co-pilot or thought partner, not the driver. And always fact check everything.

When I talk to people who don’t use AI, their reasons usually fall into two camps: they don’t trust it, or they don’t see the need. That skepticism isn’t wrong, AI does hallucinate (although less with each release) and it raises privacy questions. But the sheer scale of use suggests that ChatGPT has become a reflection of human ambition, anxiety, and curiosity all rolled into one. Six billion monthly visits tell us less about the machine farm and more about ourselves. We are asking AI to help us write, to help us think, to help us feel less alone. The open question is how we balance the gifts of velocity and nonjudgement with the costs of displacement, and whether we can design a future where those first rungs on the ladder remain intact. We cannot skip a generation in the workforce.