www.astralcodexten.com/p/ai-safety-is-not-a-corporate-plot
1 correction found
what he perceived as a lack of safety involved in scaling up ChatGPT.
This is an anachronism: Anthropic was founded in early 2021, while ChatGPT was not introduced until November 30, 2022.
Full reasoning
Anthropic’s own 2022 Series B announcement says the company was founded "at the beginning of 2021." OpenAI’s official "Introducing ChatGPT" post is dated November 30, 2022. So Dario Amodei could not have left OpenAI in 2021 because of safety concerns about scaling up ChatGPT specifically, since ChatGPT did not yet exist.
A more accurate version would be that he was concerned about OpenAI’s broader scaling direction or its large language model work, not ChatGPT by name.
2 sources
- Anthropic raises Series B to build robust AI | Anthropic
Since its founding at the beginning of 2021, Anthropic has conducted research into making systems that are more steerable, robust, and interpretable.
- Introducing ChatGPT | OpenAI
November 30, 2022 ... This post introduced ChatGPT in 2022. ... We are excited to introduce ChatGPT to get users’ feedback and learn about its strengths and weaknesses.