gvrkiran/how-people-use-chatgpt-data

Anonymized dataset on how people use ChatGPT -- obtained from 1200+ users from India, Brazil, Nigeria and Pakistan. # ChatGPT Conversation Data Across Four Countries This repository contains ChatGPT conversation data with demographic information collected from participants in four countries: **Brazil**, **India**, **Nigeria**, and **Pakistan**. Each conversation has been classified along several dimensions using LLM-based classifiers adapted from the How People Use ChatGPT (OpenAI) and the The Anthropic Economic Index report: Economic Primitives. ## Files For each country, two CSV files are provided: - `user_demographics_ .csv` — one row per participant, containing basic demographic attributes and account-type. - `user_conversations_classified_ .csv` — one row per conversation, containing conversation timestamp, classification labels, and per-conversation metadata (length, language, estimated time). | Country | # Users | # Conversations | |----------|---------|-----------------| | Brazil | 246 | 40,067 | | India | 557 | 88,958 | | Nigeria | 243 | 44,114 | | Pakistan | 206 | 29,451 | Join key between the two files: `user_id`. ## Columns: `user_demographics_ .csv` | Column | Description | |-----------|--------------------------------------------------------------------------------| | `user_id` | Anonymous participant identifier. | | `Age` | Self-reported age (years). | | `Gender` | Self-reported gender. | | `Country` | Country of residence. | | `is_plus` | Boolean. `True` if the user held a ChatGPT Plus subscription, `False` otherwise. ## Columns: `user_conversations_classified_ .csv` ### Identifiers | Column | Description | |--------------------------|---- …