ChatGPT
Era 5 · Generative AI, reasoning, and agents (2022–2026) › Frontier models
- Lineage
- Neurons and networks
- Kind
- Popularization
Who
OpenAI
What happened
OpenAI releases ChatGPT, a conversational model fine-tuned from GPT-3.5 with reinforcement learning from human feedback. It is offered free as a research preview and within two months reaches 100 million monthly users.
Why it mattered
Language models reach the general public. 100 million monthly users in two months.
The full story
Until late 2022, large language models were used mainly through programming interfaces. On November 30 OpenAI released ChatGPT as a free research preview that anyone could hold a conversation with.
According to the announcement, the dialogue format made it possible for the model to "answer followup questions, admit its mistakes, challenge incorrect premises, and reject inappropriate requests." It was a sibling model to InstructGPT, fine-tuned from a model in the GPT-3.5 series with reinforcement learning from human feedback: human trainers wrote example conversations, then ranked alternative responses from the model from best to worst, and those rankings trained a reward model that guided the fine-tuning.
The announcement itself warned of its main flaw: it sometimes wrote "plausible-sounding but incorrect or nonsensical answers."
The response was enormous. ChatGPT reached a million users in five days and, according to an estimate by the bank UBS, 100 million monthly users in two months, making it the fastest-growing internet application until Threads arrived. In December 2022 Google declared a "code red." ChatGPT Plus, a $20-a-month subscription, followed in February 2023, and in February 2026 OpenAI announced 900 million weekly users.
Builds on
Reinforcement learning from human preferences (June 12, 2017)
Documented direct influence
The same technique, by way of Ziegler (2019)
InstructGPT (March 4, 2022)
Documented direct influence
OpenAI presents ChatGPT as a sibling model to InstructGPT, trained with RLHF using the same methods
Threads that pass through here
- From reward to alignment (step 12 of 12)
Sources
Original source
Scholarship and context
Further reading
How to cite
“ChatGPT.” The History of AI, revised November 13, 2026. https://thehistoryofai.com/en/milestone/chatgpt
This entry was last reviewed on November 13, 2026.