The company behind the GPT model family and ChatGPT. Founded 2015; its GPT-3 and InstructGPT results established in-context learning and RLHF as standard practice.
OpenAI is the company behind the GPT model family and ChatGPT. Founded in 2015 as a non-profit research lab and restructured with a capped-profit arm in 2019, it published the GPT series that established decoder-only transformers as the dominant design for language models.
The 2020 paper "Language Models are Few-Shot Learners" introduced GPT-3 and the finding that scale alone produces in-context learning, so a task can be specified in the prompt instead of by fine-tuning. InstructGPT (2022) established RLHF as the standard step between a raw pre-trained model and a usable assistant. Both results are now assumed in every commercial model.
ChatGPT's browsing and search features turned a language model into a destination for queries that previously went to a search engine. OpenAI operates separate crawlers for training and for live answers, GPTBot, OAI-SearchBot and ChatGPT-User, each with its own robots.txt token. Which pages its retrieval layer returns, and which of those it cites, is one half of AI visibility.
