OpenAI is founded

OpenAI was announced on 11 December 2015 at the NeurIPS conference in Montréal, where Ilya Sutskever had been recruited from Google over a dinner with Elon Musk and Sam Altman. It was a non-profit, with a billion dollars pledged by Musk, Altman, Reid Hoffman, Peter Thiel and others, and a stated purpose of advancing digital intelligence in the way most likely to benefit humanity, "unconstrained by a need to generate financial return". Its founders said openly that the point was to ensure the technology was not controlled by one company, meaning Google, which had bought DeepMind two years before.

For its first three years it did reinforcement learning, robotics and games, beating professional Dota 2 players in 2018. Then it bet on language. GPT in 2018, GPT-2 in 2019 and GPT-3 in 2020 were each the largest language model of their moment, and the scaling-law research that justified them came from the same building.

The non-profit structure did not survive the cost of the bet. A capped-profit subsidiary was created in 2019 and Microsoft invested a billion dollars; Musk had left the board the year before and later sued. By the time ChatGPT shipped in 2022, the laboratory founded to counterbalance Google was the most valuable startup in the world.