1.5 million chats expose how people really use ChatGPT and for what reasons

Real-world use of ChatGPT revealed

OpenAI and Harvard economist David Deming released a major study of 1.5 million conversations with ChatGPT. The analysis indicates that this AI tool has moved past being a tech fad.

It is now being used for many different purposes by a much larger group worldwide, suggesting it’s becoming a key technology in daily life.

The study’s findings are significant because they show how people use ChatGPT. The data suggests it’s not just a tool for a small group of tech experts.

A growing global and diverse user base

ChatGPT’s user base is becoming much more diverse and widespread. The study found that its adoption in low-income countries grew four times faster than in high-income countries.

This rapid growth shows the tool is becoming accessible to many people worldwide, making it a truly global technology. The gender gap in usage is also closing. In January 2024, only 37% of users had typically feminine names. By July 2025, that number had grown to 52%, a percentage that mirrors the adult population.

Most chats are for everyday tasks

A surprising finding is that most people use ChatGPT for non-work-related tasks. The study shows that 70% of consumer chats are for personal use, while only 30% are for professional use.

This means the AI is helping people with things in their daily lives more than with their jobs. Even with this split, both types of usage are growing. This points to ChatGPT becoming a tool people rely on for various personal needs. It shows that the AI is just as valuable as a life assistant as a work tool.

Writing and information are top uses

Three of every four conversations with ChatGPT fall into three main categories: writing, information seeking, and practical guidance. This highlights what people value most from the AI.

For professional use, writing is the most common task. People often use the tool to draft emails or reports. Other tasks like coding and creative expression remain much smaller, or “niche,” uses for the AI.

The three ways we talk to ChatGPT

The study organized user behavior into three main types: “Asking,” “Doing,” and “Expressing.” The biggest group is Asking, making up 49% of all messages. This shows that people often use ChatGPT like an advisor, asking for information or advice.

The Doing category is next at 40%, including drafting, planning, and programming. The smallest category is Expressing, at 11%, used for personal reflection and creative play.

ChatGPT offers decision support

A key way people use ChatGPT is for “decision support.” The AI helps people by providing information and ideas to guide their thinking and streamline tasks. This is especially helpful in jobs where people must process lots of information.

The study found that using AI as an advisor or research assistant is a key way for people to get value from it. The AI isn’t just a simple tool for tasks; it helps with judgment and saves time, especially in jobs that require a lot of thinking.

Writing is a top professional use

Writing is the most popular use for work-related chats. The study shows that over two-thirds of all writing messages ask ChatGPT to edit or critique the user’s written text.

This means people use AI more to refine their words than to create new ones from scratch. This finding highlights that people see ChatGPT as a tool for improving their work. The AI is a brilliant editor and proofreader, which helps people save time and improve the quality of their professional writing.

The role of coding and self-expression

While many people think of AI for coding, the study shows a tiny part of how people use ChatGPT. Only 4.2% of all messages are related to computer programming. This contrasts with other AI models, where coding might be more popular.

Similarly, self-expression is a niche use case, at just 11% of messages. This category includes things like personal reflection and games. The data confirms that most people use the AI for practical reasons, not emotional or creative ones.

Google’s advanced AI

The AI field is competitive, with many companies creating powerful tools. Google Gemini is a major player, and it is a set of advanced AI models from Google DeepMind. Gemini is a multimodal model, meaning it can understand and work with different information types simultaneously, like text, images, and audio.

Google has released different versions of Gemini for various uses, from the most powerful Gemini Ultra, which is not yet widely available, to the more widely used Gemini Pro. This model is integrated into many Google products, like Search, making it a powerful competitor.

Microsoft’s AI assistant: Copilot

Microsoft has also entered the AI race with Copilot. This AI assistant is integrated into many Microsoft products, including Windows, Office, and GitHub. It helps users with tasks like writing documents, creating presentations, and generating computer code.

Copilot is powered by an updated version of the technology behind ChatGPT, called GPT-4, and its later versions, like GPT-4 Turbo. This integration with widely-used software makes it a very accessible and powerful tool for millions in their daily work and personal lives.

Anthropic’s conversational AI: Claude

Another major AI model is Claude, created by the company Anthropic. It’s known for its ability to have more natural, longer conversations and is designed with safety and alignment in mind. This makes it a popular choice for many businesses.

The then-latest version, Claude 3, is promoted for its reasoning and writing capabilities. Claude is a strong competitor to ChatGPT and is often used for writing, analysis, and coding in professional settings.

Meta’s integrated AI

Meta, the company behind Facebook and Instagram, has developed its own AI assistant called Meta AI. It is being integrated into messaging apps like WhatsApp and Messenger so users can access AI features inside the apps they already use.

Meta AI can answer questions, create images, and help with many other simple tasks in chat conversations. This move shows how big tech companies are trying to make AI a seamless part of our daily social and online lives.

A pioneer in AI is IBM Watson

One of the most famous early AIs was IBM Watson. It gained worldwide fame in 2011 by competing on the TV quiz show Jeopardy! and beating two of its all-time champions. Watson was a significant breakthrough in how a computer could understand human language and answer complex questions.

While Watson’s victory was groundbreaking, it used a different type of AI called symbolic AI and statistical NLP, not the transformer-based architecture used by today’s most powerful large language models.

Google’s early language breakthrough

A few years ago, Google created an important language model called BERT. It was a huge step forward in how computers understand human language.

Prior to BERT, many language models processed text in a left-to-right or right-to-left fashion; BERT introduced bidirectional encoding, considering context from both sides.

This new approach enabled the model better to understand the meaning and context of a sentence. BERT was used to improve Google Search, laying the foundation for the more advanced AI models you see today.

The first chatbot

The first chatbot was created in 1966 by a computer scientist named Joseph Weizenbaum. The program was called ELIZA and was designed to act like a therapist. It would respond to what a user said by rephrasing their words as a question.

ELIZA didn’t understand the conversations, but it was a big step. It showed that a computer could converse with a person using natural language, which was terrific then. It started the journey for all the chatbots that exist today.

Curious how Apple is stepping up its AI game? Take a look at Apple’s take on ChatGPT with Siri upgrade.

The origin of “robot” in a 1921 play

The word “robot” was first used in a play written in 1921 by a Czech author named Karel Čapek. The play was called Rossum’s Universal Robots (R.U.R.). The story was about artificial people created to do all the work for humans.

This play introduced the idea of machines that could look and act like people to do labor. It was long before computers could do anything like this, but it planted the idea in people’s minds and helped shape how we think about AI and robots today.

Want to know why privacy is becoming a hot debate? Read more in Sam Altman flags privacy risks in ChatGPT therapy.

What’s your main way of using ChatGPT? Share in the comments.

Read More From This Brand:

• ChatGPT chats found on Google — how to delete yours
• ChatGPT’s study mode could change how you learn
• OpenAI unveils ChatGPT parental controls with guidance from health experts

Don’t forget to follow us for more exclusive content right here on MSN

If you liked this story, you’ll LOVE our FREE emails. Join today and be the first to get stories like this one

This slideshow was made with AI assistance and human editing.