On November 30, 2022, OpenAI announced ChatGPT, a model trained to interact in conversation. OpenAI’s own announcement page describes the system as able to “answer followup questions, admit its mistakes, challenge incorrect premises, and reject inappropriate requests,” interacting in what it calls a conversational way.
The announcement explains that ChatGPT was trained using Reinforcement Learning from Human Feedback (RLHF), a method in which human trainers ranked model responses to align the system toward helpful dialogue. OpenAI framed the release as a research preview offered free during the testing period to gather feedback on strengths and weaknesses.
ChatGPT mattered because it put a capable large language model behind a simple chat interface that anyone could use without technical setup. It became one of the fastest-adopted consumer software products on record and triggered an industry-wide race in generative AI, prompting competing releases and large investments across the sector.
The product built directly on OpenAI’s prior GPT-3 work and the transformer architecture, packaging that research into a form business users and the general public could try immediately.