ChatGPT
- ai
OpenAI introduced ChatGPT as a public research preview on 30 November 2022. The initial service presented a GPT-3.5-family language model through a conversational interface, with training that included demonstrations and human preference feedback. Users could ask follow-up questions and request text or programming assistance in ordinary language. The launch brought large language models to a much broader public audience. It did not create the underlying Transformer architecture or eliminate the possibility of confident errors: OpenAI’s announcement explicitly described reliability limitations. This entry concerns the initial release, whose capabilities and conditions should be distinguished from later versions. [Launch announcement](https://openai.com/index/chatgpt/). **Timeline anchor:** 30 November 2022 CE. **References:** [Source 1](https://en.wikipedia.org/wiki/ChatGPT).
Sources
Related
On other timelines
- ChatGPT on The history of artificial intelligence by @athan