The call for AI safety to admit it was wrong about GPT-2
A post quotes an AI:AM guest saying he has seen no mea culpa from OpenAI, while describing him as being on the company's side.
TLDR
According to a post sharing his AI:AM remarks, a guest argued that AI safety's first step should be admitting it was wrong about GPT-2. The post contrasts Hinton owning his radiologist miss with the guest's comment about OpenAI: “I still haven't seen any mea culpa. None.” It describes the criticism as coming from someone on OpenAI's side.
Combined views
552
1 Source, first seen 15d ago